re-structure Java API, add Scala API and consolidate the names of Java/Scala API
This commit is contained in:
CodingCat
@@ -0,0 +1,152 @@
|
||||
package org.dmlc.xgboost4j;
|
||||
|
||||
import java.io.IOException;
|
||||
import java.util.Map;
|
||||
|
||||
public interface Booster {
|
||||
|
||||
/**
|
||||
* set parameter
|
||||
*
|
||||
* @param key param name
|
||||
* @param value param value
|
||||
*/
|
||||
void setParam(String key, String value) throws XGBoostError;
|
||||
|
||||
/**
|
||||
* set parameters
|
||||
*
|
||||
* @param params parameters key-value map
|
||||
*/
|
||||
void setParams(Map<String, Object> params) throws XGBoostError;
|
||||
|
||||
/**
|
||||
* Update (one iteration)
|
||||
*
|
||||
* @param dtrain training data
|
||||
* @param iter current iteration number
|
||||
*/
|
||||
void update(DMatrix dtrain, int iter) throws XGBoostError;
|
||||
|
||||
/**
|
||||
* update with customize obj func
|
||||
*
|
||||
* @param dtrain training data
|
||||
* @param obj customized objective class
|
||||
*/
|
||||
void update(DMatrix dtrain, IObjective obj) throws XGBoostError;
|
||||
|
||||
/**
|
||||
* update with give grad and hess
|
||||
*
|
||||
* @param dtrain training data
|
||||
* @param grad first order of gradient
|
||||
* @param hess seconde order of gradient
|
||||
*/
|
||||
void boost(DMatrix dtrain, float[] grad, float[] hess) throws XGBoostError;
|
||||
|
||||
/**
|
||||
* evaluate with given dmatrixs.
|
||||
*
|
||||
* @param evalMatrixs dmatrixs for evaluation
|
||||
* @param evalNames name for eval dmatrixs, used for check results
|
||||
* @param iter current eval iteration
|
||||
* @return eval information
|
||||
*/
|
||||
String evalSet(DMatrix[] evalMatrixs, String[] evalNames, int iter) throws XGBoostError;
|
||||
|
||||
/**
|
||||
* evaluate with given customized Evaluation class
|
||||
*
|
||||
* @param evalMatrixs evaluation matrix
|
||||
* @param evalNames evaluation names
|
||||
* @param eval custom evaluator
|
||||
* @return eval information
|
||||
*/
|
||||
String evalSet(DMatrix[] evalMatrixs, String[] evalNames, IEvaluation eval) throws XGBoostError;
|
||||
|
||||
/**
|
||||
* Predict with data
|
||||
*
|
||||
* @param data dmatrix storing the input
|
||||
* @return predict result
|
||||
*/
|
||||
float[][] predict(DMatrix data) throws XGBoostError;
|
||||
|
||||
|
||||
/**
|
||||
* Predict with data
|
||||
*
|
||||
* @param data dmatrix storing the input
|
||||
* @param outPutMargin Whether to output the raw untransformed margin value.
|
||||
* @return predict result
|
||||
*/
|
||||
float[][] predict(DMatrix data, boolean outPutMargin) throws XGBoostError;
|
||||
|
||||
|
||||
/**
|
||||
* Predict with data
|
||||
*
|
||||
* @param data dmatrix storing the input
|
||||
* @param outPutMargin Whether to output the raw untransformed margin value.
|
||||
* @param treeLimit Limit number of trees in the prediction; defaults to 0 (use all trees).
|
||||
* @return predict result
|
||||
*/
|
||||
float[][] predict(DMatrix data, boolean outPutMargin, int treeLimit) throws XGBoostError;
|
||||
|
||||
|
||||
/**
|
||||
* Predict with data
|
||||
* @param data dmatrix storing the input
|
||||
* @param treeLimit Limit number of trees in the prediction; defaults to 0 (use all trees).
|
||||
* @param predLeaf When this option is on, the output will be a matrix of (nsample, ntrees), nsample = data.numRow
|
||||
with each record indicating the predicted leaf index of each sample in each tree.
|
||||
Note that the leaf index of a tree is unique per tree, so you may find leaf 1
|
||||
in both tree 1 and tree 0.
|
||||
* @return predict result
|
||||
* @throws XGBoostError native error
|
||||
*/
|
||||
float[][] predict(DMatrix data, int treeLimit, boolean predLeaf) throws XGBoostError;
|
||||
|
||||
/**
|
||||
* save model to modelPath
|
||||
*
|
||||
* @param modelPath model path
|
||||
*/
|
||||
void saveModel(String modelPath) throws XGBoostError;
|
||||
|
||||
/**
|
||||
* Dump model into a text file.
|
||||
*
|
||||
* @param modelPath file to save dumped model info
|
||||
* @param withStats bool Controls whether the split statistics are output.
|
||||
*/
|
||||
void dumpModel(String modelPath, boolean withStats) throws IOException, XGBoostError;
|
||||
|
||||
/**
|
||||
* Dump model into a text file.
|
||||
*
|
||||
* @param modelPath file to save dumped model info
|
||||
* @param featureMap featureMap file
|
||||
* @param withStats bool
|
||||
* Controls whether the split statistics are output.
|
||||
*/
|
||||
void dumpModel(String modelPath, String featureMap, boolean withStats) throws IOException, XGBoostError;
|
||||
|
||||
/**
|
||||
* get importance of each feature
|
||||
*
|
||||
* @return featureMap key: feature index, value: feature importance score
|
||||
*/
|
||||
Map<String, Integer> getFeatureScore() throws XGBoostError ;
|
||||
|
||||
/**
|
||||
* get importance of each feature
|
||||
*
|
||||
* @param featureMap file to save dumped model info
|
||||
* @return featureMap key: feature index, value: feature importance score
|
||||
*/
|
||||
Map<String, Integer> getFeatureScore(String featureMap) throws XGBoostError;
|
||||
|
||||
void dispose();
|
||||
}
|
||||
@@ -0,0 +1,265 @@
|
||||
/*
|
||||
Copyright (c) 2014 by Contributors
|
||||
|
||||
Licensed under the Apache License, Version 2.0 (the "License");
|
||||
you may not use this file except in compliance with the License.
|
||||
You may obtain a copy of the License at
|
||||
|
||||
http://www.apache.org/licenses/LICENSE-2.0
|
||||
|
||||
Unless required by applicable law or agreed to in writing, software
|
||||
distributed under the License is distributed on an "AS IS" BASIS,
|
||||
WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
|
||||
See the License for the specific language governing permissions and
|
||||
limitations under the License.
|
||||
*/
|
||||
package org.dmlc.xgboost4j;
|
||||
|
||||
import org.apache.commons.logging.Log;
|
||||
import org.apache.commons.logging.LogFactory;
|
||||
|
||||
import java.io.IOException;
|
||||
|
||||
/**
|
||||
* DMatrix for xgboost, similar to the python wrapper xgboost.py
|
||||
*
|
||||
* @author hzx
|
||||
*/
|
||||
public class DMatrix {
|
||||
private static final Log logger = LogFactory.getLog(DMatrix.class);
|
||||
private long handle = 0;
|
||||
|
||||
//load native library
|
||||
static {
|
||||
try {
|
||||
NativeLibLoader.InitXgboost();
|
||||
} catch (IOException ex) {
|
||||
logger.error("load native library failed.");
|
||||
logger.error(ex);
|
||||
}
|
||||
}
|
||||
|
||||
/**
|
||||
* sparse matrix type (CSR or CSC)
|
||||
*/
|
||||
public static enum SparseType {
|
||||
CSR,
|
||||
CSC;
|
||||
}
|
||||
|
||||
public DMatrix(String dataPath) throws XGBoostError {
|
||||
if (dataPath == null) {
|
||||
throw new NullPointerException("dataPath: null");
|
||||
}
|
||||
long[] out = new long[1];
|
||||
JNIErrorHandle.checkCall(XgboostJNI.XGDMatrixCreateFromFile(dataPath, 1, out));
|
||||
handle = out[0];
|
||||
}
|
||||
|
||||
public DMatrix(long[] headers, int[] indices, float[] data, SparseType st) throws XGBoostError {
|
||||
long[] out = new long[1];
|
||||
if (st == SparseType.CSR) {
|
||||
JNIErrorHandle.checkCall(XgboostJNI.XGDMatrixCreateFromCSR(headers, indices, data, out));
|
||||
} else if (st == SparseType.CSC) {
|
||||
JNIErrorHandle.checkCall(XgboostJNI.XGDMatrixCreateFromCSC(headers, indices, data, out));
|
||||
} else {
|
||||
throw new UnknownError("unknow sparsetype");
|
||||
}
|
||||
handle = out[0];
|
||||
}
|
||||
|
||||
/**
|
||||
* create DMatrix from dense matrix
|
||||
*
|
||||
* @param data data values
|
||||
* @param nrow number of rows
|
||||
* @param ncol number of columns
|
||||
* @throws XGBoostError native error
|
||||
*/
|
||||
public DMatrix(float[] data, int nrow, int ncol) throws XGBoostError {
|
||||
long[] out = new long[1];
|
||||
JNIErrorHandle.checkCall(XgboostJNI.XGDMatrixCreateFromMat(data, nrow, ncol, 0.0f, out));
|
||||
handle = out[0];
|
||||
}
|
||||
|
||||
/**
|
||||
* used for DMatrix slice
|
||||
*
|
||||
* @param handle
|
||||
*/
|
||||
protected DMatrix(long handle) {
|
||||
this.handle = handle;
|
||||
}
|
||||
|
||||
|
||||
/**
|
||||
* set label of dmatrix
|
||||
*
|
||||
* @param labels labels
|
||||
* @throws XGBoostError native error
|
||||
*/
|
||||
public void setLabel(float[] labels) throws XGBoostError {
|
||||
JNIErrorHandle.checkCall(XgboostJNI.XGDMatrixSetFloatInfo(handle, "label", labels));
|
||||
}
|
||||
|
||||
/**
|
||||
* set weight of each instance
|
||||
*
|
||||
* @param weights weights
|
||||
* @throws XGBoostError native error
|
||||
*/
|
||||
public void setWeight(float[] weights) throws XGBoostError {
|
||||
JNIErrorHandle.checkCall(XgboostJNI.XGDMatrixSetFloatInfo(handle, "weight", weights));
|
||||
}
|
||||
|
||||
/**
|
||||
* if specified, xgboost will start from this init margin
|
||||
* can be used to specify initial prediction to boost from
|
||||
*
|
||||
* @param baseMargin base margin
|
||||
* @throws XGBoostError native error
|
||||
*/
|
||||
public void setBaseMargin(float[] baseMargin) throws XGBoostError {
|
||||
JNIErrorHandle.checkCall(XgboostJNI.XGDMatrixSetFloatInfo(handle, "base_margin", baseMargin));
|
||||
}
|
||||
|
||||
/**
|
||||
* if specified, xgboost will start from this init margin
|
||||
* can be used to specify initial prediction to boost from
|
||||
*
|
||||
* @param baseMargin base margin
|
||||
* @throws XGBoostError native error
|
||||
*/
|
||||
public void setBaseMargin(float[][] baseMargin) throws XGBoostError {
|
||||
float[] flattenMargin = flatten(baseMargin);
|
||||
setBaseMargin(flattenMargin);
|
||||
}
|
||||
|
||||
/**
|
||||
* Set group sizes of DMatrix (used for ranking)
|
||||
*
|
||||
* @param group group size as array
|
||||
* @throws XGBoostError native error
|
||||
*/
|
||||
public void setGroup(int[] group) throws XGBoostError {
|
||||
JNIErrorHandle.checkCall(XgboostJNI.XGDMatrixSetGroup(handle, group));
|
||||
}
|
||||
|
||||
private float[] getFloatInfo(String field) throws XGBoostError {
|
||||
float[][] infos = new float[1][];
|
||||
JNIErrorHandle.checkCall(XgboostJNI.XGDMatrixGetFloatInfo(handle, field, infos));
|
||||
return infos[0];
|
||||
}
|
||||
|
||||
private int[] getIntInfo(String field) throws XGBoostError {
|
||||
int[][] infos = new int[1][];
|
||||
JNIErrorHandle.checkCall(XgboostJNI.XGDMatrixGetUIntInfo(handle, field, infos));
|
||||
return infos[0];
|
||||
}
|
||||
|
||||
/**
|
||||
* get label values
|
||||
*
|
||||
* @return label
|
||||
* @throws XGBoostError native error
|
||||
*/
|
||||
public float[] getLabel() throws XGBoostError {
|
||||
return getFloatInfo("label");
|
||||
}
|
||||
|
||||
/**
|
||||
* get weight of the DMatrix
|
||||
*
|
||||
* @return weights
|
||||
* @throws XGBoostError native error
|
||||
*/
|
||||
public float[] getWeight() throws XGBoostError {
|
||||
return getFloatInfo("weight");
|
||||
}
|
||||
|
||||
/**
|
||||
* get base margin of the DMatrix
|
||||
*
|
||||
* @return base margin
|
||||
* @throws XGBoostError native error
|
||||
*/
|
||||
public float[] getBaseMargin() throws XGBoostError {
|
||||
return getFloatInfo("base_margin");
|
||||
}
|
||||
|
||||
/**
|
||||
* Slice the DMatrix and return a new DMatrix that only contains `rowIndex`.
|
||||
*
|
||||
* @param rowIndex row index
|
||||
* @return sliced new DMatrix
|
||||
* @throws XGBoostError native error
|
||||
*/
|
||||
public DMatrix slice(int[] rowIndex) throws XGBoostError {
|
||||
long[] out = new long[1];
|
||||
JNIErrorHandle.checkCall(XgboostJNI.XGDMatrixSliceDMatrix(handle, rowIndex, out));
|
||||
long sHandle = out[0];
|
||||
DMatrix sMatrix = new DMatrix(sHandle);
|
||||
return sMatrix;
|
||||
}
|
||||
|
||||
/**
|
||||
* get the row number of DMatrix
|
||||
*
|
||||
* @return number of rows
|
||||
* @throws XGBoostError native error
|
||||
*/
|
||||
public long rowNum() throws XGBoostError {
|
||||
long[] rowNum = new long[1];
|
||||
JNIErrorHandle.checkCall(XgboostJNI.XGDMatrixNumRow(handle, rowNum));
|
||||
return rowNum[0];
|
||||
}
|
||||
|
||||
/**
|
||||
* save DMatrix to filePath
|
||||
*
|
||||
* @param filePath file path
|
||||
*/
|
||||
public void saveBinary(String filePath) {
|
||||
XgboostJNI.XGDMatrixSaveBinary(handle, filePath, 1);
|
||||
}
|
||||
|
||||
/**
|
||||
* Get the handle
|
||||
*
|
||||
* @return native handler id
|
||||
*/
|
||||
public long getHandle() {
|
||||
return handle;
|
||||
}
|
||||
|
||||
/**
|
||||
* flatten a mat to array
|
||||
*
|
||||
* @param mat
|
||||
* @return
|
||||
*/
|
||||
private static float[] flatten(float[][] mat) {
|
||||
int size = 0;
|
||||
for (float[] array : mat) size += array.length;
|
||||
float[] result = new float[size];
|
||||
int pos = 0;
|
||||
for (float[] ar : mat) {
|
||||
System.arraycopy(ar, 0, result, pos, ar.length);
|
||||
pos += ar.length;
|
||||
}
|
||||
|
||||
return result;
|
||||
}
|
||||
|
||||
@Override
|
||||
protected void finalize() {
|
||||
dispose();
|
||||
}
|
||||
|
||||
public synchronized void dispose() {
|
||||
if (handle != 0) {
|
||||
XgboostJNI.XGDMatrixFree(handle);
|
||||
handle = 0;
|
||||
}
|
||||
}
|
||||
}
|
||||
@@ -0,0 +1,39 @@
|
||||
/*
|
||||
Copyright (c) 2014 by Contributors
|
||||
|
||||
Licensed under the Apache License, Version 2.0 (the "License");
|
||||
you may not use this file except in compliance with the License.
|
||||
You may obtain a copy of the License at
|
||||
|
||||
http://www.apache.org/licenses/LICENSE-2.0
|
||||
|
||||
Unless required by applicable law or agreed to in writing, software
|
||||
distributed under the License is distributed on an "AS IS" BASIS,
|
||||
WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
|
||||
See the License for the specific language governing permissions and
|
||||
limitations under the License.
|
||||
*/
|
||||
package org.dmlc.xgboost4j;
|
||||
|
||||
/**
|
||||
* interface for customized evaluation
|
||||
*
|
||||
* @author hzx
|
||||
*/
|
||||
public interface IEvaluation {
|
||||
/**
|
||||
* get evaluate metric
|
||||
*
|
||||
* @return evalMetric
|
||||
*/
|
||||
String getMetric();
|
||||
|
||||
/**
|
||||
* evaluate with predicts and data
|
||||
*
|
||||
* @param predicts predictions as array
|
||||
* @param dmat data matrix to evaluate
|
||||
* @return result of the metric
|
||||
*/
|
||||
float eval(float[][] predicts, DMatrix dmat);
|
||||
}
|
||||
@@ -0,0 +1,34 @@
|
||||
/*
|
||||
Copyright (c) 2014 by Contributors
|
||||
|
||||
Licensed under the Apache License, Version 2.0 (the "License");
|
||||
you may not use this file except in compliance with the License.
|
||||
You may obtain a copy of the License at
|
||||
|
||||
http://www.apache.org/licenses/LICENSE-2.0
|
||||
|
||||
Unless required by applicable law or agreed to in writing, software
|
||||
distributed under the License is distributed on an "AS IS" BASIS,
|
||||
WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
|
||||
See the License for the specific language governing permissions and
|
||||
limitations under the License.
|
||||
*/
|
||||
package org.dmlc.xgboost4j;
|
||||
|
||||
import java.util.List;
|
||||
|
||||
/**
|
||||
* interface for customize Object function
|
||||
*
|
||||
* @author hzx
|
||||
*/
|
||||
public interface IObjective {
|
||||
/**
|
||||
* user define objective function, return gradient and second order gradient
|
||||
*
|
||||
* @param predicts untransformed margin predicts
|
||||
* @param dtrain training data
|
||||
* @return List with two float array, correspond to first order grad and second order grad
|
||||
*/
|
||||
List<float[]> getGradient(float[][] predicts, DMatrix dtrain);
|
||||
}
|
||||
@@ -0,0 +1,51 @@
|
||||
/*
|
||||
Copyright (c) 2014 by Contributors
|
||||
|
||||
Licensed under the Apache License, Version 2.0 (the "License");
|
||||
you may not use this file except in compliance with the License.
|
||||
You may obtain a copy of the License at
|
||||
|
||||
http://www.apache.org/licenses/LICENSE-2.0
|
||||
|
||||
Unless required by applicable law or agreed to in writing, software
|
||||
distributed under the License is distributed on an "AS IS" BASIS,
|
||||
WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
|
||||
See the License for the specific language governing permissions and
|
||||
limitations under the License.
|
||||
*/
|
||||
package org.dmlc.xgboost4j;
|
||||
|
||||
import org.apache.commons.logging.Log;
|
||||
import org.apache.commons.logging.LogFactory;
|
||||
|
||||
import java.io.IOException;
|
||||
|
||||
/**
|
||||
* Error handle for Xgboost.
|
||||
*/
|
||||
class JNIErrorHandle {
|
||||
|
||||
private static final Log logger = LogFactory.getLog(DMatrix.class);
|
||||
|
||||
//load native library
|
||||
static {
|
||||
try {
|
||||
NativeLibLoader.InitXgboost();
|
||||
} catch (IOException ex) {
|
||||
logger.error("load native library failed.");
|
||||
logger.error(ex);
|
||||
}
|
||||
}
|
||||
|
||||
/**
|
||||
* Check the return value of C API.
|
||||
*
|
||||
* @param ret return valud of xgboostJNI C API call
|
||||
* @throws XGBoostError native error
|
||||
*/
|
||||
static void checkCall(int ret) throws XGBoostError {
|
||||
if (ret != 0) {
|
||||
throw new XGBoostError(XgboostJNI.XGBGetLastError());
|
||||
}
|
||||
}
|
||||
}
|
||||
@@ -0,0 +1,467 @@
|
||||
/*
|
||||
Copyright (c) 2014 by Contributors
|
||||
|
||||
Licensed under the Apache License, Version 2.0 (the "License");
|
||||
you may not use this file except in compliance with the License.
|
||||
You may obtain a copy of the License at
|
||||
|
||||
http://www.apache.org/licenses/LICENSE-2.0
|
||||
|
||||
Unless required by applicable law or agreed to in writing, software
|
||||
distributed under the License is distributed on an "AS IS" BASIS,
|
||||
WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
|
||||
See the License for the specific language governing permissions and
|
||||
limitations under the License.
|
||||
*/
|
||||
package org.dmlc.xgboost4j;
|
||||
|
||||
import org.apache.commons.logging.Log;
|
||||
import org.apache.commons.logging.LogFactory;
|
||||
|
||||
import java.io.*;
|
||||
import java.util.HashMap;
|
||||
import java.util.List;
|
||||
import java.util.Map;
|
||||
|
||||
|
||||
/**
|
||||
* Booster for xgboost, similar to the python wrapper xgboost.py
|
||||
* but custom obj function and eval function not supported at present.
|
||||
*
|
||||
* @author hzx
|
||||
*/
|
||||
class JavaBoosterImpl implements Booster {
|
||||
private static final Log logger = LogFactory.getLog(JavaBoosterImpl.class);
|
||||
|
||||
long handle = 0;
|
||||
|
||||
//load native library
|
||||
static {
|
||||
try {
|
||||
NativeLibLoader.InitXgboost();
|
||||
} catch (IOException ex) {
|
||||
logger.error("load native library failed.");
|
||||
logger.error(ex);
|
||||
}
|
||||
}
|
||||
|
||||
/**
|
||||
* init Booster from dMatrixs
|
||||
*
|
||||
* @param params parameters
|
||||
* @param dMatrixs DMatrix array
|
||||
* @throws XGBoostError native error
|
||||
*/
|
||||
JavaBoosterImpl(Map<String, Object> params, DMatrix[] dMatrixs) throws XGBoostError {
|
||||
init(dMatrixs);
|
||||
setParam("seed", "0");
|
||||
setParams(params);
|
||||
}
|
||||
|
||||
|
||||
/**
|
||||
* load model from modelPath
|
||||
*
|
||||
* @param params parameters
|
||||
* @param modelPath booster modelPath (model generated by booster.saveModel)
|
||||
* @throws XGBoostError native error
|
||||
*/
|
||||
JavaBoosterImpl(Map<String, Object> params, String modelPath) throws XGBoostError {
|
||||
init(null);
|
||||
if (modelPath == null) {
|
||||
throw new NullPointerException("modelPath : null");
|
||||
}
|
||||
loadModel(modelPath);
|
||||
setParam("seed", "0");
|
||||
setParams(params);
|
||||
}
|
||||
|
||||
|
||||
private void init(DMatrix[] dMatrixs) throws XGBoostError {
|
||||
long[] handles = null;
|
||||
if (dMatrixs != null) {
|
||||
handles = dMatrixs2handles(dMatrixs);
|
||||
}
|
||||
long[] out = new long[1];
|
||||
JNIErrorHandle.checkCall(XgboostJNI.XGBoosterCreate(handles, out));
|
||||
|
||||
handle = out[0];
|
||||
}
|
||||
|
||||
/**
|
||||
* set parameter
|
||||
*
|
||||
* @param key param name
|
||||
* @param value param value
|
||||
* @throws XGBoostError native error
|
||||
*/
|
||||
public final void setParam(String key, String value) throws XGBoostError {
|
||||
JNIErrorHandle.checkCall(XgboostJNI.XGBoosterSetParam(handle, key, value));
|
||||
}
|
||||
|
||||
/**
|
||||
* set parameters
|
||||
*
|
||||
* @param params parameters key-value map
|
||||
* @throws XGBoostError native error
|
||||
*/
|
||||
public void setParams(Map<String, Object> params) throws XGBoostError {
|
||||
if (params != null) {
|
||||
for (Map.Entry<String, Object> entry : params.entrySet()) {
|
||||
setParam(entry.getKey(), entry.getValue().toString());
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
|
||||
/**
|
||||
* Update (one iteration)
|
||||
*
|
||||
* @param dtrain training data
|
||||
* @param iter current iteration number
|
||||
* @throws XGBoostError native error
|
||||
*/
|
||||
public void update(DMatrix dtrain, int iter) throws XGBoostError {
|
||||
JNIErrorHandle.checkCall(XgboostJNI.XGBoosterUpdateOneIter(handle, iter, dtrain.getHandle()));
|
||||
}
|
||||
|
||||
/**
|
||||
* update with customize obj func
|
||||
*
|
||||
* @param dtrain training data
|
||||
* @param obj customized objective class
|
||||
* @throws XGBoostError native error
|
||||
*/
|
||||
public void update(DMatrix dtrain, IObjective obj) throws XGBoostError {
|
||||
float[][] predicts = predict(dtrain, true);
|
||||
List<float[]> gradients = obj.getGradient(predicts, dtrain);
|
||||
boost(dtrain, gradients.get(0), gradients.get(1));
|
||||
}
|
||||
|
||||
/**
|
||||
* update with give grad and hess
|
||||
*
|
||||
* @param dtrain training data
|
||||
* @param grad first order of gradient
|
||||
* @param hess seconde order of gradient
|
||||
* @throws XGBoostError native error
|
||||
*/
|
||||
public void boost(DMatrix dtrain, float[] grad, float[] hess) throws XGBoostError {
|
||||
if (grad.length != hess.length) {
|
||||
throw new AssertionError(String.format("grad/hess length mismatch %s / %s", grad.length,
|
||||
hess.length));
|
||||
}
|
||||
JNIErrorHandle.checkCall(XgboostJNI.XGBoosterBoostOneIter(handle, dtrain.getHandle(), grad, hess));
|
||||
}
|
||||
|
||||
/**
|
||||
* evaluate with given dmatrixs.
|
||||
*
|
||||
* @param evalMatrixs dmatrixs for evaluation
|
||||
* @param evalNames name for eval dmatrixs, used for check results
|
||||
* @param iter current eval iteration
|
||||
* @return eval information
|
||||
* @throws XGBoostError native error
|
||||
*/
|
||||
public String evalSet(DMatrix[] evalMatrixs, String[] evalNames, int iter) throws XGBoostError {
|
||||
long[] handles = dMatrixs2handles(evalMatrixs);
|
||||
String[] evalInfo = new String[1];
|
||||
JNIErrorHandle.checkCall(XgboostJNI.XGBoosterEvalOneIter(handle, iter, handles, evalNames, evalInfo));
|
||||
return evalInfo[0];
|
||||
}
|
||||
|
||||
/**
|
||||
* evaluate with given customized Evaluation class
|
||||
*
|
||||
* @param evalMatrixs evaluation matrix
|
||||
* @param evalNames evaluation names
|
||||
* @param eval custom evaluator
|
||||
* @return eval information
|
||||
* @throws XGBoostError native error
|
||||
*/
|
||||
public String evalSet(DMatrix[] evalMatrixs, String[] evalNames, IEvaluation eval)
|
||||
throws XGBoostError {
|
||||
String evalInfo = "";
|
||||
for (int i = 0; i < evalNames.length; i++) {
|
||||
String evalName = evalNames[i];
|
||||
DMatrix evalMat = evalMatrixs[i];
|
||||
float evalResult = eval.eval(predict(evalMat), evalMat);
|
||||
String evalMetric = eval.getMetric();
|
||||
evalInfo += String.format("\t%s-%s:%f", evalName, evalMetric, evalResult);
|
||||
}
|
||||
return evalInfo;
|
||||
}
|
||||
|
||||
/**
|
||||
* base function for Predict
|
||||
*
|
||||
* @param data data
|
||||
* @param outPutMargin output margin
|
||||
* @param treeLimit limit number of trees
|
||||
* @param predLeaf prediction minimum to keep leafs
|
||||
* @return predict results
|
||||
*/
|
||||
private synchronized float[][] pred(DMatrix data, boolean outPutMargin, int treeLimit,
|
||||
boolean predLeaf) throws XGBoostError {
|
||||
int optionMask = 0;
|
||||
if (outPutMargin) {
|
||||
optionMask = 1;
|
||||
}
|
||||
if (predLeaf) {
|
||||
optionMask = 2;
|
||||
}
|
||||
float[][] rawPredicts = new float[1][];
|
||||
JNIErrorHandle.checkCall(XgboostJNI.XGBoosterPredict(handle, data.getHandle(), optionMask,
|
||||
treeLimit, rawPredicts));
|
||||
int row = (int) data.rowNum();
|
||||
int col = rawPredicts[0].length / row;
|
||||
float[][] predicts = new float[row][col];
|
||||
int r, c;
|
||||
for (int i = 0; i < rawPredicts[0].length; i++) {
|
||||
r = i / col;
|
||||
c = i % col;
|
||||
predicts[r][c] = rawPredicts[0][i];
|
||||
}
|
||||
return predicts;
|
||||
}
|
||||
|
||||
/**
|
||||
* Predict with data
|
||||
*
|
||||
* @param data dmatrix storing the input
|
||||
* @return predict result
|
||||
* @throws XGBoostError native error
|
||||
*/
|
||||
public float[][] predict(DMatrix data) throws XGBoostError {
|
||||
return pred(data, false, 0, false);
|
||||
}
|
||||
|
||||
/**
|
||||
* Predict with data
|
||||
*
|
||||
* @param data dmatrix storing the input
|
||||
* @param outPutMargin Whether to output the raw untransformed margin value.
|
||||
* @return predict result
|
||||
* @throws XGBoostError native error
|
||||
*/
|
||||
public float[][] predict(DMatrix data, boolean outPutMargin) throws XGBoostError {
|
||||
return pred(data, outPutMargin, 0, false);
|
||||
}
|
||||
|
||||
/**
|
||||
* Predict with data
|
||||
*
|
||||
* @param data dmatrix storing the input
|
||||
* @param outPutMargin Whether to output the raw untransformed margin value.
|
||||
* @param treeLimit Limit number of trees in the prediction; defaults to 0 (use all trees).
|
||||
* @return predict result
|
||||
* @throws XGBoostError native error
|
||||
*/
|
||||
public float[][] predict(DMatrix data, boolean outPutMargin, int treeLimit) throws XGBoostError {
|
||||
return pred(data, outPutMargin, treeLimit, false);
|
||||
}
|
||||
|
||||
/**
|
||||
* Predict with data
|
||||
*
|
||||
* @param data dmatrix storing the input
|
||||
* @param treeLimit Limit number of trees in the prediction; defaults to 0 (use all trees).
|
||||
* @param predLeaf When this option is on, the output will be a matrix of (nsample, ntrees),
|
||||
* nsample = data.numRow with each record indicating the predicted leaf index
|
||||
* of each sample in each tree.
|
||||
* Note that the leaf index of a tree is unique per tree, so you may find leaf 1
|
||||
* in both tree 1 and tree 0.
|
||||
* @return predict result
|
||||
* @throws XGBoostError native error
|
||||
*/
|
||||
public float[][] predict(DMatrix data, int treeLimit, boolean predLeaf) throws XGBoostError {
|
||||
return pred(data, false, treeLimit, predLeaf);
|
||||
}
|
||||
|
||||
/**
|
||||
* save model to modelPath
|
||||
*
|
||||
* @param modelPath model path
|
||||
*/
|
||||
public void saveModel(String modelPath) throws XGBoostError{
|
||||
JNIErrorHandle.checkCall(XgboostJNI.XGBoosterSaveModel(handle, modelPath));
|
||||
}
|
||||
|
||||
private void loadModel(String modelPath) {
|
||||
XgboostJNI.XGBoosterLoadModel(handle, modelPath);
|
||||
}
|
||||
|
||||
/**
|
||||
* get the dump of the model as a string array
|
||||
*
|
||||
* @param withStats Controls whether the split statistics are output.
|
||||
* @return dumped model information
|
||||
* @throws XGBoostError native error
|
||||
*/
|
||||
private String[] getDumpInfo(boolean withStats) throws XGBoostError {
|
||||
int statsFlag = 0;
|
||||
if (withStats) {
|
||||
statsFlag = 1;
|
||||
}
|
||||
String[][] modelInfos = new String[1][];
|
||||
JNIErrorHandle.checkCall(XgboostJNI.XGBoosterDumpModel(handle, "", statsFlag, modelInfos));
|
||||
return modelInfos[0];
|
||||
}
|
||||
|
||||
/**
|
||||
* get the dump of the model as a string array
|
||||
*
|
||||
* @param featureMap featureMap file
|
||||
* @param withStats Controls whether the split statistics are output.
|
||||
* @return dumped model information
|
||||
* @throws XGBoostError native error
|
||||
*/
|
||||
private String[] getDumpInfo(String featureMap, boolean withStats) throws XGBoostError {
|
||||
int statsFlag = 0;
|
||||
if (withStats) {
|
||||
statsFlag = 1;
|
||||
}
|
||||
String[][] modelInfos = new String[1][];
|
||||
JNIErrorHandle.checkCall(XgboostJNI.XGBoosterDumpModel(handle, featureMap, statsFlag, modelInfos));
|
||||
return modelInfos[0];
|
||||
}
|
||||
|
||||
/**
|
||||
* Dump model into a text file.
|
||||
*
|
||||
* @param modelPath file to save dumped model info
|
||||
* @param withStats bool
|
||||
* Controls whether the split statistics are output.
|
||||
* @throws FileNotFoundException file not found
|
||||
* @throws UnsupportedEncodingException unsupported feature
|
||||
* @throws IOException error with model writing
|
||||
* @throws XGBoostError native error
|
||||
*/
|
||||
public void dumpModel(String modelPath, boolean withStats) throws IOException, XGBoostError {
|
||||
File tf = new File(modelPath);
|
||||
FileOutputStream out = new FileOutputStream(tf);
|
||||
BufferedWriter writer = new BufferedWriter(new OutputStreamWriter(out, "UTF-8"));
|
||||
String[] modelInfos = getDumpInfo(withStats);
|
||||
|
||||
for (int i = 0; i < modelInfos.length; i++) {
|
||||
writer.write("booster [" + i + "]:\n");
|
||||
writer.write(modelInfos[i]);
|
||||
}
|
||||
|
||||
writer.close();
|
||||
out.close();
|
||||
}
|
||||
|
||||
|
||||
/**
|
||||
* Dump model into a text file.
|
||||
*
|
||||
* @param modelPath file to save dumped model info
|
||||
* @param featureMap featureMap file
|
||||
* @param withStats bool
|
||||
* Controls whether the split statistics are output.
|
||||
* @throws FileNotFoundException exception
|
||||
* @throws UnsupportedEncodingException exception
|
||||
* @throws IOException exception
|
||||
* @throws XGBoostError native error
|
||||
*/
|
||||
public void dumpModel(String modelPath, String featureMap, boolean withStats) throws
|
||||
IOException, XGBoostError {
|
||||
File tf = new File(modelPath);
|
||||
FileOutputStream out = new FileOutputStream(tf);
|
||||
BufferedWriter writer = new BufferedWriter(new OutputStreamWriter(out, "UTF-8"));
|
||||
String[] modelInfos = getDumpInfo(featureMap, withStats);
|
||||
|
||||
for (int i = 0; i < modelInfos.length; i++) {
|
||||
writer.write("booster [" + i + "]:\n");
|
||||
writer.write(modelInfos[i]);
|
||||
}
|
||||
|
||||
writer.close();
|
||||
out.close();
|
||||
}
|
||||
|
||||
|
||||
/**
|
||||
* get importance of each feature
|
||||
*
|
||||
* @return featureMap key: feature index, value: feature importance score
|
||||
* @throws XGBoostError native error
|
||||
*/
|
||||
public Map<String, Integer> getFeatureScore() throws XGBoostError {
|
||||
String[] modelInfos = getDumpInfo(false);
|
||||
Map<String, Integer> featureScore = new HashMap<String, Integer>();
|
||||
for (String tree : modelInfos) {
|
||||
for (String node : tree.split("\n")) {
|
||||
String[] array = node.split("\\[");
|
||||
if (array.length == 1) {
|
||||
continue;
|
||||
}
|
||||
String fid = array[1].split("\\]")[0];
|
||||
fid = fid.split("<")[0];
|
||||
if (featureScore.containsKey(fid)) {
|
||||
featureScore.put(fid, 1 + featureScore.get(fid));
|
||||
} else {
|
||||
featureScore.put(fid, 1);
|
||||
}
|
||||
}
|
||||
}
|
||||
return featureScore;
|
||||
}
|
||||
|
||||
|
||||
/**
|
||||
* get importance of each feature
|
||||
*
|
||||
* @param featureMap file to save dumped model info
|
||||
* @return featureMap key: feature index, value: feature importance score
|
||||
* @throws XGBoostError native error
|
||||
*/
|
||||
public Map<String, Integer> getFeatureScore(String featureMap) throws XGBoostError {
|
||||
String[] modelInfos = getDumpInfo(featureMap, false);
|
||||
Map<String, Integer> featureScore = new HashMap<String, Integer>();
|
||||
for (String tree : modelInfos) {
|
||||
for (String node : tree.split("\n")) {
|
||||
String[] array = node.split("\\[");
|
||||
if (array.length == 1) {
|
||||
continue;
|
||||
}
|
||||
String fid = array[1].split("\\]")[0];
|
||||
fid = fid.split("<")[0];
|
||||
if (featureScore.containsKey(fid)) {
|
||||
featureScore.put(fid, 1 + featureScore.get(fid));
|
||||
} else {
|
||||
featureScore.put(fid, 1);
|
||||
}
|
||||
}
|
||||
}
|
||||
return featureScore;
|
||||
}
|
||||
|
||||
/**
|
||||
* transfer DMatrix array to handle array (used for native functions)
|
||||
*
|
||||
* @param dmatrixs
|
||||
* @return handle array for input dmatrixs
|
||||
*/
|
||||
private static long[] dMatrixs2handles(DMatrix[] dmatrixs) {
|
||||
long[] handles = new long[dmatrixs.length];
|
||||
for (int i = 0; i < dmatrixs.length; i++) {
|
||||
handles[i] = dmatrixs[i].getHandle();
|
||||
}
|
||||
return handles;
|
||||
}
|
||||
|
||||
@Override
|
||||
protected void finalize() throws Throwable {
|
||||
super.finalize();
|
||||
dispose();
|
||||
}
|
||||
|
||||
public synchronized void dispose() {
|
||||
if (handle != 0L) {
|
||||
XgboostJNI.XGBoosterFree(handle);
|
||||
handle = 0;
|
||||
}
|
||||
}
|
||||
}
|
||||
@@ -0,0 +1,168 @@
|
||||
/*
|
||||
Copyright (c) 2014 by Contributors
|
||||
|
||||
Licensed under the Apache License, Version 2.0 (the "License");
|
||||
you may not use this file except in compliance with the License.
|
||||
You may obtain a copy of the License at
|
||||
|
||||
http://www.apache.org/licenses/LICENSE-2.0
|
||||
|
||||
Unless required by applicable law or agreed to in writing, software
|
||||
distributed under the License is distributed on an "AS IS" BASIS,
|
||||
WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
|
||||
See the License for the specific language governing permissions and
|
||||
limitations under the License.
|
||||
*/
|
||||
package org.dmlc.xgboost4j;
|
||||
|
||||
import org.apache.commons.logging.Log;
|
||||
import org.apache.commons.logging.LogFactory;
|
||||
|
||||
import java.io.*;
|
||||
import java.lang.reflect.Field;
|
||||
|
||||
/**
|
||||
* class to load native library
|
||||
*
|
||||
* @author hzx
|
||||
*/
|
||||
class NativeLibLoader {
|
||||
private static final Log logger = LogFactory.getLog(NativeLibLoader.class);
|
||||
|
||||
private static boolean initialized = false;
|
||||
private static final String nativePath = "../lib/";
|
||||
private static final String nativeResourcePath = "/lib/";
|
||||
private static final String[] libNames = new String[]{"xgboost4j"};
|
||||
|
||||
public static synchronized void InitXgboost() throws IOException {
|
||||
if (!initialized) {
|
||||
for (String libName : libNames) {
|
||||
smartLoad(libName);
|
||||
}
|
||||
initialized = true;
|
||||
}
|
||||
}
|
||||
|
||||
/**
|
||||
* Loads library from current JAR archive
|
||||
* <p/>
|
||||
* The file from JAR is copied into system temporary directory and then loaded.
|
||||
* The temporary file is deleted after exiting.
|
||||
* Method uses String as filename because the pathname is "abstract", not system-dependent.
|
||||
* <p/>
|
||||
* The restrictions of {@link File#createTempFile(java.lang.String, java.lang.String)} apply to {@code path}.
|
||||
*
|
||||
* @param path The filename inside JAR as absolute path (beginning with '/'), e.g. /package/File.ext
|
||||
* @throws IOException If temporary file creation or read/write operation fails
|
||||
* @throws IllegalArgumentException If source file (param path) does not exist
|
||||
* @throws IllegalArgumentException If the path is not absolute or if the filename is shorter than three characters
|
||||
*/
|
||||
private static void loadLibraryFromJar(String path) throws IOException {
|
||||
|
||||
if (!path.startsWith("/")) {
|
||||
throw new IllegalArgumentException("The path has to be absolute (start with '/').");
|
||||
}
|
||||
|
||||
// Obtain filename from path
|
||||
String[] parts = path.split("/");
|
||||
String filename = (parts.length > 1) ? parts[parts.length - 1] : null;
|
||||
|
||||
// Split filename to prexif and suffix (extension)
|
||||
String prefix = "";
|
||||
String suffix = null;
|
||||
if (filename != null) {
|
||||
parts = filename.split("\\.", 2);
|
||||
prefix = parts[0];
|
||||
suffix = (parts.length > 1) ? "." + parts[parts.length - 1] : null; // Thanks, davs! :-)
|
||||
}
|
||||
|
||||
// Check if the filename is okay
|
||||
if (filename == null || prefix.length() < 3) {
|
||||
throw new IllegalArgumentException("The filename has to be at least 3 characters long.");
|
||||
}
|
||||
|
||||
// Prepare temporary file
|
||||
File temp = File.createTempFile(prefix, suffix);
|
||||
temp.deleteOnExit();
|
||||
|
||||
if (!temp.exists()) {
|
||||
throw new FileNotFoundException("File " + temp.getAbsolutePath() + " does not exist.");
|
||||
}
|
||||
|
||||
// Prepare buffer for data copying
|
||||
byte[] buffer = new byte[1024];
|
||||
int readBytes;
|
||||
|
||||
// Open and check input stream
|
||||
InputStream is = NativeLibLoader.class.getResourceAsStream(path);
|
||||
if (is == null) {
|
||||
throw new FileNotFoundException("File " + path + " was not found inside JAR.");
|
||||
}
|
||||
|
||||
// Open output stream and copy data between source file in JAR and the temporary file
|
||||
OutputStream os = new FileOutputStream(temp);
|
||||
try {
|
||||
while ((readBytes = is.read(buffer)) != -1) {
|
||||
os.write(buffer, 0, readBytes);
|
||||
}
|
||||
} finally {
|
||||
// If read/write fails, close streams safely before throwing an exception
|
||||
os.close();
|
||||
is.close();
|
||||
}
|
||||
|
||||
// Finally, load the library
|
||||
System.load(temp.getAbsolutePath());
|
||||
}
|
||||
|
||||
/**
|
||||
* load native library, this method will first try to load library from java.library.path, then
|
||||
* try to load library in jar package.
|
||||
*
|
||||
* @param libName library path
|
||||
* @throws IOException exception
|
||||
*/
|
||||
private static void smartLoad(String libName) throws IOException {
|
||||
addNativeDir(nativePath);
|
||||
try {
|
||||
System.loadLibrary(libName);
|
||||
System.out.println("======load " + libName + " successfully");
|
||||
} catch (UnsatisfiedLinkError e) {
|
||||
try {
|
||||
String libraryFromJar = nativeResourcePath + System.mapLibraryName(libName);
|
||||
loadLibraryFromJar(libraryFromJar);
|
||||
} catch (IOException e1) {
|
||||
throw e1;
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
/**
|
||||
* Add libPath to java.library.path, then native library in libPath would be load properly
|
||||
*
|
||||
* @param libPath library path
|
||||
* @throws IOException exception
|
||||
*/
|
||||
private static void addNativeDir(String libPath) throws IOException {
|
||||
try {
|
||||
Field field = ClassLoader.class.getDeclaredField("usr_paths");
|
||||
field.setAccessible(true);
|
||||
String[] paths = (String[]) field.get(null);
|
||||
for (String path : paths) {
|
||||
if (libPath.equals(path)) {
|
||||
return;
|
||||
}
|
||||
}
|
||||
String[] tmp = new String[paths.length + 1];
|
||||
System.arraycopy(paths, 0, tmp, 0, paths.length);
|
||||
tmp[paths.length] = libPath;
|
||||
field.set(null, tmp);
|
||||
} catch (IllegalAccessException e) {
|
||||
logger.error(e.getMessage());
|
||||
throw new IOException("Failed to get permissions to set library path");
|
||||
} catch (NoSuchFieldException e) {
|
||||
logger.error(e.getMessage());
|
||||
throw new IOException("Failed to get field handle to set library path");
|
||||
}
|
||||
}
|
||||
}
|
||||
@@ -0,0 +1,336 @@
|
||||
/*
|
||||
Copyright (c) 2014 by Contributors
|
||||
|
||||
Licensed under the Apache License, Version 2.0 (the "License");
|
||||
you may not use this file except in compliance with the License.
|
||||
You may obtain a copy of the License at
|
||||
|
||||
http://www.apache.org/licenses/LICENSE-2.0
|
||||
|
||||
Unless required by applicable law or agreed to in writing, software
|
||||
distributed under the License is distributed on an "AS IS" BASIS,
|
||||
WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
|
||||
See the License for the specific language governing permissions and
|
||||
limitations under the License.
|
||||
*/
|
||||
package org.dmlc.xgboost4j;
|
||||
|
||||
import org.apache.commons.logging.Log;
|
||||
import org.apache.commons.logging.LogFactory;
|
||||
|
||||
import java.util.*;
|
||||
|
||||
|
||||
/**
|
||||
* trainer for xgboost
|
||||
*
|
||||
* @author hzx
|
||||
*/
|
||||
public class XGBoost {
|
||||
private static final Log logger = LogFactory.getLog(XGBoost.class);
|
||||
|
||||
/**
|
||||
* Train a booster with given parameters.
|
||||
*
|
||||
* @param params Booster params.
|
||||
* @param dtrain Data to be trained.
|
||||
* @param round Number of boosting iterations.
|
||||
* @param watches a group of items to be evaluated during training, this allows user to watch
|
||||
* performance on the validation set.
|
||||
* @param obj customized objective (set to null if not used)
|
||||
* @param eval customized evaluation (set to null if not used)
|
||||
* @return trained booster
|
||||
* @throws XGBoostError native error
|
||||
*/
|
||||
public static Booster train(Map<String, Object> params, DMatrix dtrain, int round,
|
||||
Map<String, DMatrix> watches, IObjective obj,
|
||||
IEvaluation eval) throws XGBoostError {
|
||||
|
||||
//collect eval matrixs
|
||||
String[] evalNames;
|
||||
DMatrix[] evalMats;
|
||||
List<String> names = new ArrayList<String>();
|
||||
List<DMatrix> mats = new ArrayList<DMatrix>();
|
||||
|
||||
for (Map.Entry<String, DMatrix> evalEntry : watches.entrySet()) {
|
||||
names.add(evalEntry.getKey());
|
||||
mats.add(evalEntry.getValue());
|
||||
}
|
||||
|
||||
evalNames = names.toArray(new String[names.size()]);
|
||||
evalMats = mats.toArray(new DMatrix[mats.size()]);
|
||||
|
||||
//collect all data matrixs
|
||||
DMatrix[] allMats;
|
||||
if (evalMats != null && evalMats.length > 0) {
|
||||
allMats = new DMatrix[evalMats.length + 1];
|
||||
allMats[0] = dtrain;
|
||||
System.arraycopy(evalMats, 0, allMats, 1, evalMats.length);
|
||||
} else {
|
||||
allMats = new DMatrix[1];
|
||||
allMats[0] = dtrain;
|
||||
}
|
||||
|
||||
//initialize booster
|
||||
Booster booster = new JavaBoosterImpl(params, allMats);
|
||||
|
||||
//begin to train
|
||||
for (int iter = 0; iter < round; iter++) {
|
||||
if (obj != null) {
|
||||
booster.update(dtrain, obj);
|
||||
} else {
|
||||
booster.update(dtrain, iter);
|
||||
}
|
||||
|
||||
//evaluation
|
||||
if (evalMats != null && evalMats.length > 0) {
|
||||
String evalInfo;
|
||||
if (eval != null) {
|
||||
evalInfo = booster.evalSet(evalMats, evalNames, eval);
|
||||
} else {
|
||||
evalInfo = booster.evalSet(evalMats, evalNames, iter);
|
||||
}
|
||||
logger.info(evalInfo);
|
||||
}
|
||||
}
|
||||
return booster;
|
||||
}
|
||||
|
||||
/**
|
||||
* init Booster from dMatrixs
|
||||
*
|
||||
* @param params parameters
|
||||
* @param dMatrixs DMatrix array
|
||||
* @throws XGBoostError native error
|
||||
*/
|
||||
public static Booster initBoostingModel(
|
||||
Map<String, Object> params,
|
||||
DMatrix[] dMatrixs) throws XGBoostError {
|
||||
return new JavaBoosterImpl(params, dMatrixs);
|
||||
}
|
||||
|
||||
/**
|
||||
* load model from modelPath
|
||||
*
|
||||
* @param params parameters
|
||||
* @param modelPath booster modelPath (model generated by booster.saveModel)
|
||||
* @throws XGBoostError native error
|
||||
*/
|
||||
public static Booster loadBoostModel(Map<String, Object> params, String modelPath)
|
||||
throws XGBoostError {
|
||||
return new JavaBoosterImpl(params, modelPath);
|
||||
}
|
||||
|
||||
/**
|
||||
* Cross-validation with given paramaters.
|
||||
*
|
||||
* @param params Booster params.
|
||||
* @param data Data to be trained.
|
||||
* @param round Number of boosting iterations.
|
||||
* @param nfold Number of folds in CV.
|
||||
* @param metrics Evaluation metrics to be watched in CV.
|
||||
* @param obj customized objective (set to null if not used)
|
||||
* @param eval customized evaluation (set to null if not used)
|
||||
* @return evaluation history
|
||||
* @throws XGBoostError native error
|
||||
*/
|
||||
public static String[] crossValiation(
|
||||
Map<String, Object> params,
|
||||
DMatrix data,
|
||||
int round,
|
||||
int nfold,
|
||||
String[] metrics,
|
||||
IObjective obj,
|
||||
IEvaluation eval) throws XGBoostError {
|
||||
CVPack[] cvPacks = makeNFold(data, nfold, params, metrics);
|
||||
String[] evalHist = new String[round];
|
||||
String[] results = new String[cvPacks.length];
|
||||
for (int i = 0; i < round; i++) {
|
||||
for (CVPack cvPack : cvPacks) {
|
||||
if (obj != null) {
|
||||
cvPack.update(obj);
|
||||
} else {
|
||||
cvPack.update(i);
|
||||
}
|
||||
}
|
||||
|
||||
for (int j = 0; j < cvPacks.length; j++) {
|
||||
if (eval != null) {
|
||||
results[j] = cvPacks[j].eval(eval);
|
||||
} else {
|
||||
results[j] = cvPacks[j].eval(i);
|
||||
}
|
||||
}
|
||||
|
||||
evalHist[i] = aggCVResults(results);
|
||||
logger.info(evalHist[i]);
|
||||
}
|
||||
return evalHist;
|
||||
}
|
||||
|
||||
/**
|
||||
* make an n-fold array of CVPack from random indices
|
||||
*
|
||||
* @param data original data
|
||||
* @param nfold num of folds
|
||||
* @param params booster parameters
|
||||
* @param evalMetrics Evaluation metrics
|
||||
* @return CV package array
|
||||
* @throws XGBoostError native error
|
||||
*/
|
||||
private static CVPack[] makeNFold(DMatrix data, int nfold, Map<String, Object> params,
|
||||
String[] evalMetrics) throws XGBoostError {
|
||||
List<Integer> samples = genRandPermutationNums(0, (int) data.rowNum());
|
||||
int step = samples.size() / nfold;
|
||||
int[] testSlice = new int[step];
|
||||
int[] trainSlice = new int[samples.size() - step];
|
||||
int testid, trainid;
|
||||
CVPack[] cvPacks = new CVPack[nfold];
|
||||
for (int i = 0; i < nfold; i++) {
|
||||
testid = 0;
|
||||
trainid = 0;
|
||||
for (int j = 0; j < samples.size(); j++) {
|
||||
if (j > (i * step) && j < (i * step + step) && testid < step) {
|
||||
testSlice[testid] = samples.get(j);
|
||||
testid++;
|
||||
} else {
|
||||
if (trainid < samples.size() - step) {
|
||||
trainSlice[trainid] = samples.get(j);
|
||||
trainid++;
|
||||
} else {
|
||||
testSlice[testid] = samples.get(j);
|
||||
testid++;
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
DMatrix dtrain = data.slice(trainSlice);
|
||||
DMatrix dtest = data.slice(testSlice);
|
||||
CVPack cvPack = new CVPack(dtrain, dtest, params);
|
||||
//set eval types
|
||||
if (evalMetrics != null) {
|
||||
for (String type : evalMetrics) {
|
||||
cvPack.booster.setParam("eval_metric", type);
|
||||
}
|
||||
}
|
||||
cvPacks[i] = cvPack;
|
||||
}
|
||||
|
||||
return cvPacks;
|
||||
}
|
||||
|
||||
private static List<Integer> genRandPermutationNums(int start, int end) {
|
||||
List<Integer> samples = new ArrayList<Integer>();
|
||||
for (int i = start; i < end; i++) {
|
||||
samples.add(i);
|
||||
}
|
||||
Collections.shuffle(samples);
|
||||
return samples;
|
||||
}
|
||||
|
||||
/**
|
||||
* Aggregate cross-validation results.
|
||||
*
|
||||
* @param results eval info from each data sample
|
||||
* @return cross-validation eval info
|
||||
*/
|
||||
private static String aggCVResults(String[] results) {
|
||||
Map<String, List<Float>> cvMap = new HashMap<String, List<Float>>();
|
||||
String aggResult = results[0].split("\t")[0];
|
||||
for (String result : results) {
|
||||
String[] items = result.split("\t");
|
||||
for (int i = 1; i < items.length; i++) {
|
||||
String[] tup = items[i].split(":");
|
||||
String key = tup[0];
|
||||
Float value = Float.valueOf(tup[1]);
|
||||
if (!cvMap.containsKey(key)) {
|
||||
cvMap.put(key, new ArrayList<Float>());
|
||||
}
|
||||
cvMap.get(key).add(value);
|
||||
}
|
||||
}
|
||||
|
||||
for (String key : cvMap.keySet()) {
|
||||
float value = 0f;
|
||||
for (Float tvalue : cvMap.get(key)) {
|
||||
value += tvalue;
|
||||
}
|
||||
value /= cvMap.get(key).size();
|
||||
aggResult += String.format("\tcv-%s:%f", key, value);
|
||||
}
|
||||
|
||||
return aggResult;
|
||||
}
|
||||
|
||||
/**
|
||||
* cross validation package for xgb
|
||||
*
|
||||
* @author hzx
|
||||
*/
|
||||
private static class CVPack {
|
||||
DMatrix dtrain;
|
||||
DMatrix dtest;
|
||||
DMatrix[] dmats;
|
||||
String[] names;
|
||||
Booster booster;
|
||||
|
||||
/**
|
||||
* create an cross validation package
|
||||
*
|
||||
* @param dtrain train data
|
||||
* @param dtest test data
|
||||
* @param params parameters
|
||||
* @throws XGBoostError native error
|
||||
*/
|
||||
public CVPack(DMatrix dtrain, DMatrix dtest, Map<String, Object> params)
|
||||
throws XGBoostError {
|
||||
dmats = new DMatrix[]{dtrain, dtest};
|
||||
booster = XGBoost.initBoostingModel(params, dmats);
|
||||
names = new String[]{"train", "test"};
|
||||
this.dtrain = dtrain;
|
||||
this.dtest = dtest;
|
||||
}
|
||||
|
||||
/**
|
||||
* update one iteration
|
||||
*
|
||||
* @param iter iteration num
|
||||
* @throws XGBoostError native error
|
||||
*/
|
||||
public void update(int iter) throws XGBoostError {
|
||||
booster.update(dtrain, iter);
|
||||
}
|
||||
|
||||
/**
|
||||
* update one iteration
|
||||
*
|
||||
* @param obj customized objective
|
||||
* @throws XGBoostError native error
|
||||
*/
|
||||
public void update(IObjective obj) throws XGBoostError {
|
||||
booster.update(dtrain, obj);
|
||||
}
|
||||
|
||||
/**
|
||||
* evaluation
|
||||
*
|
||||
* @param iter iteration num
|
||||
* @return evaluation
|
||||
* @throws XGBoostError native error
|
||||
*/
|
||||
public String eval(int iter) throws XGBoostError {
|
||||
return booster.evalSet(dmats, names, iter);
|
||||
}
|
||||
|
||||
/**
|
||||
* evaluation
|
||||
*
|
||||
* @param eval customized eval
|
||||
* @return evaluation
|
||||
* @throws XGBoostError native error
|
||||
*/
|
||||
public String eval(IEvaluation eval) throws XGBoostError {
|
||||
return booster.evalSet(dmats, names, eval);
|
||||
}
|
||||
}
|
||||
}
|
||||
@@ -0,0 +1,27 @@
|
||||
/*
|
||||
Copyright (c) 2014 by Contributors
|
||||
|
||||
Licensed under the Apache License, Version 2.0 (the "License");
|
||||
you may not use this file except in compliance with the License.
|
||||
You may obtain a copy of the License at
|
||||
|
||||
http://www.apache.org/licenses/LICENSE-2.0
|
||||
|
||||
Unless required by applicable law or agreed to in writing, software
|
||||
distributed under the License is distributed on an "AS IS" BASIS,
|
||||
WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
|
||||
See the License for the specific language governing permissions and
|
||||
limitations under the License.
|
||||
*/
|
||||
package org.dmlc.xgboost4j;
|
||||
|
||||
/**
|
||||
* custom error class for xgboost
|
||||
*
|
||||
* @author hzx
|
||||
*/
|
||||
public class XGBoostError extends Exception {
|
||||
public XGBoostError(String message) {
|
||||
super(message);
|
||||
}
|
||||
}
|
||||
@@ -0,0 +1,83 @@
|
||||
/*
|
||||
Copyright (c) 2014 by Contributors
|
||||
|
||||
Licensed under the Apache License, Version 2.0 (the "License");
|
||||
you may not use this file except in compliance with the License.
|
||||
You may obtain a copy of the License at
|
||||
|
||||
http://www.apache.org/licenses/LICENSE-2.0
|
||||
|
||||
Unless required by applicable law or agreed to in writing, software
|
||||
distributed under the License is distributed on an "AS IS" BASIS,
|
||||
WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
|
||||
See the License for the specific language governing permissions and
|
||||
limitations under the License.
|
||||
*/
|
||||
package org.dmlc.xgboost4j;
|
||||
|
||||
/**
|
||||
* xgboost jni wrapper functions for xgboost_wrapper.h
|
||||
* change 2015-7-6: *use a long[] (length=1) as container of handle to get the output DMatrix or Booster
|
||||
*
|
||||
* @author hzx
|
||||
*/
|
||||
class XgboostJNI {
|
||||
public final static native String XGBGetLastError();
|
||||
|
||||
public final static native int XGDMatrixCreateFromFile(String fname, int silent, long[] out);
|
||||
|
||||
public final static native int XGDMatrixCreateFromCSR(long[] indptr, int[] indices, float[] data,
|
||||
long[] out);
|
||||
|
||||
public final static native int XGDMatrixCreateFromCSC(long[] colptr, int[] indices, float[] data,
|
||||
long[] out);
|
||||
|
||||
public final static native int XGDMatrixCreateFromMat(float[] data, int nrow, int ncol,
|
||||
float missing, long[] out);
|
||||
|
||||
public final static native int XGDMatrixSliceDMatrix(long handle, int[] idxset, long[] out);
|
||||
|
||||
public final static native int XGDMatrixFree(long handle);
|
||||
|
||||
public final static native int XGDMatrixSaveBinary(long handle, String fname, int silent);
|
||||
|
||||
public final static native int XGDMatrixSetFloatInfo(long handle, String field, float[] array);
|
||||
|
||||
public final static native int XGDMatrixSetUIntInfo(long handle, String field, int[] array);
|
||||
|
||||
public final static native int XGDMatrixSetGroup(long handle, int[] group);
|
||||
|
||||
public final static native int XGDMatrixGetFloatInfo(long handle, String field, float[][] info);
|
||||
|
||||
public final static native int XGDMatrixGetUIntInfo(long handle, String filed, int[][] info);
|
||||
|
||||
public final static native int XGDMatrixNumRow(long handle, long[] row);
|
||||
|
||||
public final static native int XGBoosterCreate(long[] handles, long[] out);
|
||||
|
||||
public final static native int XGBoosterFree(long handle);
|
||||
|
||||
public final static native int XGBoosterSetParam(long handle, String name, String value);
|
||||
|
||||
public final static native int XGBoosterUpdateOneIter(long handle, int iter, long dtrain);
|
||||
|
||||
public final static native int XGBoosterBoostOneIter(long handle, long dtrain, float[] grad,
|
||||
float[] hess);
|
||||
|
||||
public final static native int XGBoosterEvalOneIter(long handle, int iter, long[] dmats,
|
||||
String[] evnames, String[] eval_info);
|
||||
|
||||
public final static native int XGBoosterPredict(long handle, long dmat, int option_mask,
|
||||
int ntree_limit, float[][] predicts);
|
||||
|
||||
public final static native int XGBoosterLoadModel(long handle, String fname);
|
||||
|
||||
public final static native int XGBoosterSaveModel(long handle, String fname);
|
||||
|
||||
public final static native int XGBoosterLoadModelFromBuffer(long handle, long buf, long len);
|
||||
|
||||
public final static native int XGBoosterGetModelRaw(long handle, String[] out_string);
|
||||
|
||||
public final static native int XGBoosterDumpModel(long handle, String fmap, int with_stats,
|
||||
String[][] out_strings);
|
||||
}
|
||||
@@ -0,0 +1,172 @@
|
||||
package org.dmlc.xgboost4j.scala
|
||||
|
||||
import java.io.IOException
|
||||
|
||||
import scala.collection.mutable
|
||||
|
||||
import org.dmlc.xgboost4j.{IEvaluation, IObjective, XGBoostError}
|
||||
|
||||
trait Booster {
|
||||
|
||||
|
||||
/**
|
||||
* set parameter
|
||||
*
|
||||
* @param key param name
|
||||
* @param value param value
|
||||
*/
|
||||
@throws(classOf[XGBoostError])
|
||||
def setParam(key: String, value: String)
|
||||
|
||||
/**
|
||||
* set parameters
|
||||
*
|
||||
* @param params parameters key-value map
|
||||
*/
|
||||
@throws(classOf[XGBoostError])
|
||||
def setParams(params: Map[String, AnyRef])
|
||||
|
||||
/**
|
||||
* Update (one iteration)
|
||||
*
|
||||
* @param dtrain training data
|
||||
* @param iter current iteration number
|
||||
*/
|
||||
@throws(classOf[XGBoostError])
|
||||
def update(dtrain: DMatrix, iter: Int)
|
||||
|
||||
/**
|
||||
* update with customize obj func
|
||||
*
|
||||
* @param dtrain training data
|
||||
* @param obj customized objective class
|
||||
*/
|
||||
@throws(classOf[XGBoostError])
|
||||
def update(dtrain: DMatrix, obj: IObjective)
|
||||
|
||||
/**
|
||||
* update with give grad and hess
|
||||
*
|
||||
* @param dtrain training data
|
||||
* @param grad first order of gradient
|
||||
* @param hess seconde order of gradient
|
||||
*/
|
||||
@throws(classOf[XGBoostError])
|
||||
def boost(dtrain: DMatrix, grad: Array[Float], hess: Array[Float])
|
||||
|
||||
/**
|
||||
* evaluate with given dmatrixs.
|
||||
*
|
||||
* @param evalMatrixs dmatrixs for evaluation
|
||||
* @param evalNames name for eval dmatrixs, used for check results
|
||||
* @param iter current eval iteration
|
||||
* @return eval information
|
||||
*/
|
||||
@throws(classOf[XGBoostError])
|
||||
def evalSet(evalMatrixs: Array[DMatrix], evalNames: Array[String], iter: Int): String
|
||||
|
||||
/**
|
||||
* evaluate with given customized Evaluation class
|
||||
*
|
||||
* @param evalMatrixs evaluation matrix
|
||||
* @param evalNames evaluation names
|
||||
* @param eval custom evaluator
|
||||
* @return eval information
|
||||
*/
|
||||
@throws(classOf[XGBoostError])
|
||||
def evalSet(evalMatrixs: Array[DMatrix], evalNames: Array[String], eval: IEvaluation): String
|
||||
|
||||
/**
|
||||
* Predict with data
|
||||
*
|
||||
* @param data dmatrix storing the input
|
||||
* @return predict result
|
||||
*/
|
||||
@throws(classOf[XGBoostError])
|
||||
def predict(data: DMatrix): Array[Array[Float]]
|
||||
|
||||
/**
|
||||
* Predict with data
|
||||
*
|
||||
* @param data dmatrix storing the input
|
||||
* @param outPutMargin Whether to output the raw untransformed margin value.
|
||||
* @return predict result
|
||||
*/
|
||||
@throws(classOf[XGBoostError])
|
||||
def predict(data: DMatrix, outPutMargin: Boolean): Array[Array[Float]]
|
||||
|
||||
/**
|
||||
* Predict with data
|
||||
*
|
||||
* @param data dmatrix storing the input
|
||||
* @param outPutMargin Whether to output the raw untransformed margin value.
|
||||
* @param treeLimit Limit number of trees in the prediction; defaults to 0 (use all trees).
|
||||
* @return predict result
|
||||
*/
|
||||
@throws(classOf[XGBoostError])
|
||||
def predict(data: DMatrix, outPutMargin: Boolean, treeLimit: Int): Array[Array[Float]]
|
||||
|
||||
/**
|
||||
* Predict with data
|
||||
*
|
||||
* @param data dmatrix storing the input
|
||||
* @param treeLimit Limit number of trees in the prediction; defaults to 0 (use all trees).
|
||||
* @param predLeaf When this option is on, the output will be a matrix of (nsample, ntrees), nsample = data.numRow
|
||||
with each record indicating the predicted leaf index of each sample in each tree.
|
||||
Note that the leaf index of a tree is unique per tree, so you may find leaf 1
|
||||
in both tree 1 and tree 0.
|
||||
* @return predict result
|
||||
* @throws XGBoostError native error
|
||||
*/
|
||||
@throws(classOf[XGBoostError])
|
||||
def predict(data: DMatrix, treeLimit: Int, predLeaf: Boolean): Array[Array[Float]]
|
||||
|
||||
/**
|
||||
* save model to modelPath
|
||||
*
|
||||
* @param modelPath model path
|
||||
*/
|
||||
@throws(classOf[XGBoostError])
|
||||
def saveModel(modelPath: String)
|
||||
|
||||
/**
|
||||
* Dump model into a text file.
|
||||
*
|
||||
* @param modelPath file to save dumped model info
|
||||
* @param withStats bool Controls whether the split statistics are output.
|
||||
*/
|
||||
@throws(classOf[IOException])
|
||||
@throws(classOf[XGBoostError])
|
||||
def dumpModel(modelPath: String, withStats: Boolean)
|
||||
|
||||
/**
|
||||
* Dump model into a text file.
|
||||
*
|
||||
* @param modelPath file to save dumped model info
|
||||
* @param featureMap featureMap file
|
||||
* @param withStats bool
|
||||
* Controls whether the split statistics are output.
|
||||
*/
|
||||
@throws(classOf[IOException])
|
||||
@throws(classOf[XGBoostError])
|
||||
def dumpModel(modelPath: String, featureMap: String, withStats: Boolean)
|
||||
|
||||
/**
|
||||
* get importance of each feature
|
||||
*
|
||||
* @return featureMap key: feature index, value: feature importance score
|
||||
*/
|
||||
@throws(classOf[XGBoostError])
|
||||
def getFeatureScore: mutable.Map[String, Integer]
|
||||
|
||||
/**
|
||||
* get importance of each feature
|
||||
*
|
||||
* @param featureMap file to save dumped model info
|
||||
* @return featureMap key: feature index, value: feature importance score
|
||||
*/
|
||||
@throws(classOf[XGBoostError])
|
||||
def getFeatureScore(featureMap: String): mutable.Map[String, Integer]
|
||||
|
||||
def dispose
|
||||
}
|
||||
@@ -0,0 +1,161 @@
|
||||
package org.dmlc.xgboost4j.scala
|
||||
|
||||
import org.dmlc.xgboost4j.{DMatrix => JDMatrix, XGBoostError}
|
||||
|
||||
class DMatrix private(private[scala] val jDMatrix: JDMatrix) {
|
||||
|
||||
/**
|
||||
* init DMatrix from file (svmlight format)
|
||||
*
|
||||
* @param dataPath path of data file
|
||||
* @throws XGBoostError native error
|
||||
*/
|
||||
def this(dataPath: String) {
|
||||
this(new JDMatrix(dataPath))
|
||||
}
|
||||
|
||||
/**
|
||||
* create DMatrix from sparse matrix
|
||||
*
|
||||
* @param headers index to headers (rowHeaders for CSR or colHeaders for CSC)
|
||||
* @param indices Indices (colIndexs for CSR or rowIndexs for CSC)
|
||||
* @param data non zero values (sequence by row for CSR or by col for CSC)
|
||||
* @param st sparse matrix type (CSR or CSC)
|
||||
*/
|
||||
@throws(classOf[XGBoostError])
|
||||
def this(headers: Array[Long], indices: Array[Int], data: Array[Float], st: JDMatrix.SparseType) {
|
||||
this(new JDMatrix(headers, indices, data, st))
|
||||
}
|
||||
|
||||
/**
|
||||
* create DMatrix from dense matrix
|
||||
*
|
||||
* @param data data values
|
||||
* @param nrow number of rows
|
||||
* @param ncol number of columns
|
||||
*/
|
||||
@throws(classOf[XGBoostError])
|
||||
def this(data: Array[Float], nrow: Int, ncol: Int) {
|
||||
this(new JDMatrix(data, nrow, ncol))
|
||||
}
|
||||
|
||||
/**
|
||||
* set label of dmatrix
|
||||
*
|
||||
* @param labels labels
|
||||
*/
|
||||
@throws(classOf[XGBoostError])
|
||||
def setLabel(labels: Array[Float]): Unit = {
|
||||
jDMatrix.setLabel(labels)
|
||||
}
|
||||
|
||||
/**
|
||||
* set weight of each instance
|
||||
*
|
||||
* @param weights weights
|
||||
*/
|
||||
@throws(classOf[XGBoostError])
|
||||
def setWeight(weights: Array[Float]): Unit = {
|
||||
jDMatrix.setWeight(weights)
|
||||
}
|
||||
|
||||
/**
|
||||
* if specified, xgboost will start from this init margin
|
||||
* can be used to specify initial prediction to boost from
|
||||
*
|
||||
* @param baseMargin base margin
|
||||
*/
|
||||
@throws(classOf[XGBoostError])
|
||||
def setBaseMargin(baseMargin: Array[Float]): Unit = {
|
||||
jDMatrix.setBaseMargin(baseMargin)
|
||||
}
|
||||
|
||||
/**
|
||||
* if specified, xgboost will start from this init margin
|
||||
* can be used to specify initial prediction to boost from
|
||||
*
|
||||
* @param baseMargin base margin
|
||||
*/
|
||||
@throws(classOf[XGBoostError])
|
||||
def setBaseMargin(baseMargin: Array[Array[Float]]): Unit = {
|
||||
jDMatrix.setBaseMargin(baseMargin)
|
||||
}
|
||||
|
||||
/**
|
||||
* Set group sizes of DMatrix (used for ranking)
|
||||
*
|
||||
* @param group group size as array
|
||||
*/
|
||||
@throws(classOf[XGBoostError])
|
||||
def setGroup(group: Array[Int]): Unit = {
|
||||
jDMatrix.setGroup(group)
|
||||
}
|
||||
|
||||
/**
|
||||
* get label values
|
||||
*
|
||||
* @return label
|
||||
*/
|
||||
@throws(classOf[XGBoostError])
|
||||
def getLabel: Array[Float] = {
|
||||
jDMatrix.getLabel
|
||||
}
|
||||
|
||||
/**
|
||||
* get weight of the DMatrix
|
||||
*
|
||||
* @return weights
|
||||
*/
|
||||
@throws(classOf[XGBoostError])
|
||||
def getWeight: Array[Float] = {
|
||||
jDMatrix.getWeight
|
||||
}
|
||||
|
||||
/**
|
||||
* get base margin of the DMatrix
|
||||
*
|
||||
* @return base margin
|
||||
*/
|
||||
@throws(classOf[XGBoostError])
|
||||
def getBaseMargin: Array[Float] = {
|
||||
jDMatrix.getBaseMargin
|
||||
}
|
||||
|
||||
/**
|
||||
* Slice the DMatrix and return a new DMatrix that only contains `rowIndex`.
|
||||
*
|
||||
* @param rowIndex row index
|
||||
* @return sliced new DMatrix
|
||||
*/
|
||||
@throws(classOf[XGBoostError])
|
||||
def slice(rowIndex: Array[Int]): DMatrix = {
|
||||
new DMatrix(jDMatrix.slice(rowIndex))
|
||||
}
|
||||
|
||||
/**
|
||||
* get the row number of DMatrix
|
||||
*
|
||||
* @return number of rows
|
||||
*/
|
||||
@throws(classOf[XGBoostError])
|
||||
def rowNum: Long = {
|
||||
jDMatrix.rowNum
|
||||
}
|
||||
|
||||
/**
|
||||
* save DMatrix to filePath
|
||||
*
|
||||
* @param filePath file path
|
||||
*/
|
||||
def saveBinary(filePath: String): Unit = {
|
||||
jDMatrix.saveBinary(filePath)
|
||||
}
|
||||
|
||||
def getHandle: Long = {
|
||||
jDMatrix.getHandle
|
||||
}
|
||||
|
||||
def delete(): Unit = {
|
||||
jDMatrix.dispose()
|
||||
}
|
||||
}
|
||||
@@ -0,0 +1,82 @@
|
||||
package org.dmlc.xgboost4j.scala
|
||||
|
||||
import scala.collection.JavaConverters._
|
||||
import scala.collection.mutable
|
||||
|
||||
import org.dmlc.xgboost4j.{Booster => JBooster, IEvaluation, IObjective}
|
||||
|
||||
private[scala] class ScalaBoosterImpl private[xgboost4j](booster: JBooster) extends Booster {
|
||||
|
||||
override def setParam(key: String, value: String): Unit = {
|
||||
booster.setParam(key, value)
|
||||
}
|
||||
|
||||
override def update(dtrain: DMatrix, iter: Int): Unit = {
|
||||
booster.update(dtrain.jDMatrix, iter)
|
||||
}
|
||||
|
||||
override def update(dtrain: DMatrix, obj: IObjective): Unit = {
|
||||
booster.update(dtrain.jDMatrix, obj)
|
||||
}
|
||||
|
||||
override def dumpModel(modelPath: String, withStats: Boolean): Unit = {
|
||||
booster.dumpModel(modelPath, withStats)
|
||||
}
|
||||
|
||||
override def dumpModel(modelPath: String, featureMap: String, withStats: Boolean): Unit = {
|
||||
booster.dumpModel(modelPath, featureMap, withStats)
|
||||
}
|
||||
|
||||
override def setParams(params: Map[String, AnyRef]): Unit = {
|
||||
booster.setParams(params.asJava)
|
||||
}
|
||||
|
||||
override def evalSet(evalMatrixs: Array[DMatrix], evalNames: Array[String], iter: Int): String = {
|
||||
booster.evalSet(evalMatrixs.map(_.jDMatrix), evalNames, iter)
|
||||
}
|
||||
|
||||
override def evalSet(evalMatrixs: Array[DMatrix], evalNames: Array[String], eval: IEvaluation): String = {
|
||||
booster.evalSet(evalMatrixs.map(_.jDMatrix), evalNames, eval)
|
||||
}
|
||||
|
||||
override def dispose: Unit = {
|
||||
booster.dispose()
|
||||
}
|
||||
|
||||
override def predict(data: DMatrix): Array[Array[Float]] = {
|
||||
booster.predict(data.jDMatrix)
|
||||
}
|
||||
|
||||
override def predict(data: DMatrix, outPutMargin: Boolean): Array[Array[Float]] = {
|
||||
booster.predict(data.jDMatrix, outPutMargin)
|
||||
}
|
||||
|
||||
override def predict(data: DMatrix, outPutMargin: Boolean, treeLimit: Int): Array[Array[Float]] = {
|
||||
booster.predict(data.jDMatrix, outPutMargin, treeLimit)
|
||||
}
|
||||
|
||||
override def predict(data: DMatrix, treeLimit: Int, predLeaf: Boolean): Array[Array[Float]] = {
|
||||
booster.predict(data.jDMatrix, treeLimit, predLeaf)
|
||||
}
|
||||
|
||||
override def boost(dtrain: DMatrix, grad: Array[Float], hess: Array[Float]): Unit = {
|
||||
booster.boost(dtrain.jDMatrix, grad, hess)
|
||||
}
|
||||
|
||||
override def getFeatureScore: mutable.Map[String, Integer] = {
|
||||
booster.getFeatureScore.asScala
|
||||
}
|
||||
|
||||
override def getFeatureScore(featureMap: String): mutable.Map[String, Integer] = {
|
||||
booster.getFeatureScore(featureMap).asScala
|
||||
}
|
||||
|
||||
override def saveModel(modelPath: String): Unit = {
|
||||
booster.saveModel(modelPath)
|
||||
}
|
||||
|
||||
override def finalize(): Unit = {
|
||||
super.finalize()
|
||||
dispose
|
||||
}
|
||||
}
|
||||
@@ -0,0 +1,36 @@
|
||||
package org.dmlc.xgboost4j.scala
|
||||
|
||||
import _root_.scala.collection.JavaConverters._
|
||||
|
||||
import org.dmlc.xgboost4j
|
||||
import org.dmlc.xgboost4j.{XGBoost => JXGBoost, IEvaluation, IObjective}
|
||||
|
||||
object XGBoost {
|
||||
|
||||
def train(params: Map[String, AnyRef], dtrain: xgboost4j.DMatrix, round: Int,
|
||||
watches: Map[String, xgboost4j.DMatrix], obj: IObjective, eval: IEvaluation): Booster = {
|
||||
val xgboostInJava = JXGBoost.train(params.asJava, dtrain, round, watches.asJava, obj, eval)
|
||||
new ScalaBoosterImpl(xgboostInJava)
|
||||
}
|
||||
|
||||
def crossValiation(params: Map[String, AnyRef],
|
||||
data: DMatrix,
|
||||
round: Int,
|
||||
nfold: Int,
|
||||
metrics: Array[String],
|
||||
obj: IObjective,
|
||||
eval: IEvaluation): Array[String] = {
|
||||
JXGBoost.crossValiation(params.asJava, data.jDMatrix, round, nfold, metrics, obj,
|
||||
eval)
|
||||
}
|
||||
|
||||
def initBoostModel(params: Map[String, AnyRef], dMatrixs: Array[DMatrix]): Booster = {
|
||||
val xgboostInJava = JXGBoost.initBoostingModel(params.asJava, dMatrixs.map(_.jDMatrix))
|
||||
new ScalaBoosterImpl(xgboostInJava)
|
||||
}
|
||||
|
||||
def loadBoostModel(params: Map[String, AnyRef], modelPath: String): Booster = {
|
||||
val xgboostInJava = JXGBoost.loadBoostModel(params.asJava, modelPath)
|
||||
new ScalaBoosterImpl(xgboostInJava)
|
||||
}
|
||||
}
|
||||
513
jvm-packages/xgboost4j/src/native/xgboost4j.cpp
Normal file
513
jvm-packages/xgboost4j/src/native/xgboost4j.cpp
Normal file
@@ -0,0 +1,513 @@
|
||||
/*
|
||||
Copyright (c) 2014 by Contributors
|
||||
Licensed under the Apache License, Version 2.0 (the "License");
|
||||
you may not use this file except in compliance with the License.
|
||||
You may obtain a copy of the License at
|
||||
|
||||
http://www.apache.org/licenses/LICENSE-2.0
|
||||
Unless required by applicable law or agreed to in writing, software
|
||||
distributed under the License is distributed on an "AS IS" BASIS,
|
||||
WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
|
||||
See the License for the specific language governing permissions and
|
||||
limitations under the License.
|
||||
*/
|
||||
|
||||
#include "xgboost/c_api.h"
|
||||
#include "xgboost4j.h"
|
||||
#include <cstring>
|
||||
|
||||
//helper functions
|
||||
//set handle
|
||||
void setHandle(JNIEnv *jenv, jlongArray jhandle, void* handle) {
|
||||
long out[1];
|
||||
out[0] = (long) handle;
|
||||
jenv->SetLongArrayRegion(jhandle, 0, 1, (const jlong*) out);
|
||||
}
|
||||
|
||||
JNIEXPORT jstring JNICALL Java_org_dmlc_xgboost4j_XgboostJNI_XGBGetLastError
|
||||
(JNIEnv *jenv, jclass jcls) {
|
||||
jstring jresult = 0 ;
|
||||
const char* result = XGBGetLastError();
|
||||
if (result) jresult = jenv->NewStringUTF(result);
|
||||
return jresult;
|
||||
}
|
||||
|
||||
JNIEXPORT jint JNICALL Java_org_dmlc_xgboost4j_XgboostJNI_XGDMatrixCreateFromFile
|
||||
(JNIEnv *jenv, jclass jcls, jstring jfname, jint jsilent, jlongArray jout) {
|
||||
DMatrixHandle result;
|
||||
const char* fname = jenv->GetStringUTFChars(jfname, 0);
|
||||
int ret = XGDMatrixCreateFromFile(fname, jsilent, &result);
|
||||
if (fname) jenv->ReleaseStringUTFChars(jfname, fname);
|
||||
setHandle(jenv, jout, result);
|
||||
return ret;
|
||||
}
|
||||
|
||||
/*
|
||||
* Class: org_dmlc_xgboost4j_XgboostJNI
|
||||
* Method: XGDMatrixCreateFromCSR
|
||||
* Signature: ([J[J[F)J
|
||||
*/
|
||||
JNIEXPORT jint JNICALL Java_org_dmlc_xgboost4j_XgboostJNI_XGDMatrixCreateFromCSR
|
||||
(JNIEnv *jenv, jclass jcls, jlongArray jindptr, jintArray jindices, jfloatArray jdata, jlongArray jout) {
|
||||
DMatrixHandle result;
|
||||
jlong* indptr = jenv->GetLongArrayElements(jindptr, 0);
|
||||
jint* indices = jenv->GetIntArrayElements(jindices, 0);
|
||||
jfloat* data = jenv->GetFloatArrayElements(jdata, 0);
|
||||
bst_ulong nindptr = (bst_ulong)jenv->GetArrayLength(jindptr);
|
||||
bst_ulong nelem = (bst_ulong)jenv->GetArrayLength(jdata);
|
||||
int ret = (jint) XGDMatrixCreateFromCSR((unsigned long const *)indptr, (unsigned int const *)indices, (float const *)data, nindptr, nelem, &result);
|
||||
setHandle(jenv, jout, result);
|
||||
//Release
|
||||
jenv->ReleaseLongArrayElements(jindptr, indptr, 0);
|
||||
jenv->ReleaseIntArrayElements(jindices, indices, 0);
|
||||
jenv->ReleaseFloatArrayElements(jdata, data, 0);
|
||||
return ret;
|
||||
}
|
||||
|
||||
/*
|
||||
* Class: org_dmlc_xgboost4j_XgboostJNI
|
||||
* Method: XGDMatrixCreateFromCSC
|
||||
* Signature: ([J[J[F)J
|
||||
*/
|
||||
JNIEXPORT jint JNICALL Java_org_dmlc_xgboost4j_XgboostJNI_XGDMatrixCreateFromCSC
|
||||
(JNIEnv *jenv, jclass jcls, jlongArray jindptr, jintArray jindices, jfloatArray jdata, jlongArray jout) {
|
||||
DMatrixHandle result;
|
||||
jlong* indptr = jenv->GetLongArrayElements(jindptr, NULL);
|
||||
jint* indices = jenv->GetIntArrayElements(jindices, 0);
|
||||
jfloat* data = jenv->GetFloatArrayElements(jdata, NULL);
|
||||
bst_ulong nindptr = (bst_ulong)jenv->GetArrayLength(jindptr);
|
||||
bst_ulong nelem = (bst_ulong)jenv->GetArrayLength(jdata);
|
||||
|
||||
int ret = (jint) XGDMatrixCreateFromCSC((unsigned long const *)indptr, (unsigned int const *)indices, (float const *)data, nindptr, nelem, &result);
|
||||
setHandle(jenv, jout, result);
|
||||
//release
|
||||
jenv->ReleaseLongArrayElements(jindptr, indptr, 0);
|
||||
jenv->ReleaseIntArrayElements(jindices, indices, 0);
|
||||
jenv->ReleaseFloatArrayElements(jdata, data, 0);
|
||||
|
||||
return ret;
|
||||
}
|
||||
|
||||
/*
|
||||
* Class: org_dmlc_xgboost4j_XgboostJNI
|
||||
* Method: XGDMatrixCreateFromMat
|
||||
* Signature: ([FIIF)J
|
||||
*/
|
||||
JNIEXPORT jint JNICALL Java_org_dmlc_xgboost4j_XgboostJNI_XGDMatrixCreateFromMat
|
||||
(JNIEnv *jenv, jclass jcls, jfloatArray jdata, jint jnrow, jint jncol, jfloat jmiss, jlongArray jout) {
|
||||
DMatrixHandle result;
|
||||
jfloat* data = jenv->GetFloatArrayElements(jdata, 0);
|
||||
bst_ulong nrow = (bst_ulong)jnrow;
|
||||
bst_ulong ncol = (bst_ulong)jncol;
|
||||
int ret = (jint) XGDMatrixCreateFromMat((float const *)data, nrow, ncol, jmiss, &result);
|
||||
setHandle(jenv, jout, result);
|
||||
//release
|
||||
jenv->ReleaseFloatArrayElements(jdata, data, 0);
|
||||
return ret;
|
||||
}
|
||||
|
||||
/*
|
||||
* Class: org_dmlc_xgboost4j_XgboostJNI
|
||||
* Method: XGDMatrixSliceDMatrix
|
||||
* Signature: (J[I)J
|
||||
*/
|
||||
JNIEXPORT jint JNICALL Java_org_dmlc_xgboost4j_XgboostJNI_XGDMatrixSliceDMatrix
|
||||
(JNIEnv *jenv, jclass jcls, jlong jhandle, jintArray jindexset, jlongArray jout) {
|
||||
DMatrixHandle result;
|
||||
DMatrixHandle handle = (DMatrixHandle) jhandle;
|
||||
|
||||
jint* indexset = jenv->GetIntArrayElements(jindexset, 0);
|
||||
bst_ulong len = (bst_ulong)jenv->GetArrayLength(jindexset);
|
||||
|
||||
int ret = XGDMatrixSliceDMatrix(handle, (int const *)indexset, len, &result);
|
||||
setHandle(jenv, jout, result);
|
||||
//release
|
||||
jenv->ReleaseIntArrayElements(jindexset, indexset, 0);
|
||||
|
||||
return ret;
|
||||
}
|
||||
|
||||
/*
|
||||
* Class: org_dmlc_xgboost4j_XgboostJNI
|
||||
* Method: XGDMatrixFree
|
||||
* Signature: (J)V
|
||||
*/
|
||||
JNIEXPORT jint JNICALL Java_org_dmlc_xgboost4j_XgboostJNI_XGDMatrixFree
|
||||
(JNIEnv *jenv, jclass jcls, jlong jhandle) {
|
||||
DMatrixHandle handle = (DMatrixHandle) jhandle;
|
||||
int ret = XGDMatrixFree(handle);
|
||||
return ret;
|
||||
}
|
||||
|
||||
/*
|
||||
* Class: org_dmlc_xgboost4j_XgboostJNI
|
||||
* Method: XGDMatrixSaveBinary
|
||||
* Signature: (JLjava/lang/String;I)V
|
||||
*/
|
||||
JNIEXPORT jint JNICALL Java_org_dmlc_xgboost4j_XgboostJNI_XGDMatrixSaveBinary
|
||||
(JNIEnv *jenv, jclass jcls, jlong jhandle, jstring jfname, jint jsilent) {
|
||||
DMatrixHandle handle = (DMatrixHandle) jhandle;
|
||||
const char* fname = jenv->GetStringUTFChars(jfname, 0);
|
||||
int ret = XGDMatrixSaveBinary(handle, fname, jsilent);
|
||||
if (fname) jenv->ReleaseStringUTFChars(jfname, (const char *)fname);
|
||||
return ret;
|
||||
}
|
||||
|
||||
/*
|
||||
* Class: org_dmlc_xgboost4j_XgboostJNI
|
||||
* Method: XGDMatrixSetFloatInfo
|
||||
* Signature: (JLjava/lang/String;[F)V
|
||||
*/
|
||||
JNIEXPORT jint JNICALL Java_org_dmlc_xgboost4j_XgboostJNI_XGDMatrixSetFloatInfo
|
||||
(JNIEnv *jenv, jclass jcls, jlong jhandle, jstring jfield, jfloatArray jarray) {
|
||||
DMatrixHandle handle = (DMatrixHandle) jhandle;
|
||||
const char* field = jenv->GetStringUTFChars(jfield, 0);
|
||||
|
||||
jfloat* array = jenv->GetFloatArrayElements(jarray, NULL);
|
||||
bst_ulong len = (bst_ulong)jenv->GetArrayLength(jarray);
|
||||
int ret = XGDMatrixSetFloatInfo(handle, field, (float const *)array, len);
|
||||
//release
|
||||
if (field) jenv->ReleaseStringUTFChars(jfield, field);
|
||||
jenv->ReleaseFloatArrayElements(jarray, array, 0);
|
||||
return ret;
|
||||
}
|
||||
|
||||
/*
|
||||
* Class: org_dmlc_xgboost4j_XgboostJNI
|
||||
* Method: XGDMatrixSetUIntInfo
|
||||
* Signature: (JLjava/lang/String;[I)V
|
||||
*/
|
||||
JNIEXPORT jint JNICALL Java_org_dmlc_xgboost4j_XgboostJNI_XGDMatrixSetUIntInfo
|
||||
(JNIEnv *jenv, jclass jcls, jlong jhandle, jstring jfield, jintArray jarray) {
|
||||
DMatrixHandle handle = (DMatrixHandle) jhandle;
|
||||
const char* field = jenv->GetStringUTFChars(jfield, 0);
|
||||
jint* array = jenv->GetIntArrayElements(jarray, NULL);
|
||||
bst_ulong len = (bst_ulong)jenv->GetArrayLength(jarray);
|
||||
int ret = XGDMatrixSetUIntInfo(handle, (char const *)field, (unsigned int const *)array, len);
|
||||
//release
|
||||
if (field) jenv->ReleaseStringUTFChars(jfield, (const char *)field);
|
||||
jenv->ReleaseIntArrayElements(jarray, array, 0);
|
||||
|
||||
return ret;
|
||||
}
|
||||
|
||||
/*
|
||||
* Class: org_dmlc_xgboost4j_XgboostJNI
|
||||
* Method: XGDMatrixSetGroup
|
||||
* Signature: (J[I)V
|
||||
*/
|
||||
JNIEXPORT jint JNICALL Java_org_dmlc_xgboost4j_XgboostJNI_XGDMatrixSetGroup
|
||||
(JNIEnv * jenv, jclass jcls, jlong jhandle, jintArray jarray) {
|
||||
DMatrixHandle handle = (DMatrixHandle) jhandle;
|
||||
jint* array = jenv->GetIntArrayElements(jarray, NULL);
|
||||
bst_ulong len = (bst_ulong)jenv->GetArrayLength(jarray);
|
||||
int ret = XGDMatrixSetGroup(handle, (unsigned int const *)array, len);
|
||||
//release
|
||||
jenv->ReleaseIntArrayElements(jarray, array, 0);
|
||||
return ret;
|
||||
}
|
||||
|
||||
/*
|
||||
* Class: org_dmlc_xgboost4j_XgboostJNI
|
||||
* Method: XGDMatrixGetFloatInfo
|
||||
* Signature: (JLjava/lang/String;)[F
|
||||
*/
|
||||
JNIEXPORT jint JNICALL Java_org_dmlc_xgboost4j_XgboostJNI_XGDMatrixGetFloatInfo
|
||||
(JNIEnv *jenv, jclass jcls, jlong jhandle, jstring jfield, jobjectArray jout) {
|
||||
DMatrixHandle handle = (DMatrixHandle) jhandle;
|
||||
const char* field = jenv->GetStringUTFChars(jfield, 0);
|
||||
bst_ulong len;
|
||||
float *result;
|
||||
int ret = XGDMatrixGetFloatInfo(handle, field, &len, (const float**) &result);
|
||||
if (field) jenv->ReleaseStringUTFChars(jfield, field);
|
||||
|
||||
jsize jlen = (jsize) len;
|
||||
jfloatArray jarray = jenv->NewFloatArray(jlen);
|
||||
jenv->SetFloatArrayRegion(jarray, 0, jlen, (jfloat *) result);
|
||||
jenv->SetObjectArrayElement(jout, 0, (jobject) jarray);
|
||||
|
||||
return ret;
|
||||
}
|
||||
|
||||
/*
|
||||
* Class: org_dmlc_xgboost4j_XgboostJNI
|
||||
* Method: XGDMatrixGetUIntInfo
|
||||
* Signature: (JLjava/lang/String;)[I
|
||||
*/
|
||||
JNIEXPORT jint JNICALL Java_org_dmlc_xgboost4j_XgboostJNI_XGDMatrixGetUIntInfo
|
||||
(JNIEnv *jenv, jclass jcls, jlong jhandle, jstring jfield, jobjectArray jout) {
|
||||
DMatrixHandle handle = (DMatrixHandle) jhandle;
|
||||
const char* field = jenv->GetStringUTFChars(jfield, 0);
|
||||
bst_ulong len;
|
||||
unsigned int *result;
|
||||
int ret = (jint) XGDMatrixGetUIntInfo(handle, field, &len, (const unsigned int **) &result);
|
||||
if (field) jenv->ReleaseStringUTFChars(jfield, field);
|
||||
|
||||
jsize jlen = (jsize) len;
|
||||
jintArray jarray = jenv->NewIntArray(jlen);
|
||||
jenv->SetIntArrayRegion(jarray, 0, jlen, (jint *) result);
|
||||
jenv->SetObjectArrayElement(jout, 0, jarray);
|
||||
return ret;
|
||||
}
|
||||
|
||||
/*
|
||||
* Class: org_dmlc_xgboost4j_XgboostJNI
|
||||
* Method: XGDMatrixNumRow
|
||||
* Signature: (J)J
|
||||
*/
|
||||
JNIEXPORT jint JNICALL Java_org_dmlc_xgboost4j_XgboostJNI_XGDMatrixNumRow
|
||||
(JNIEnv *jenv, jclass jcls, jlong jhandle, jlongArray jout) {
|
||||
DMatrixHandle handle = (DMatrixHandle) jhandle;
|
||||
bst_ulong result[1];
|
||||
int ret = (jint) XGDMatrixNumRow(handle, result);
|
||||
jenv->SetLongArrayRegion(jout, 0, 1, (const jlong *) result);
|
||||
return ret;
|
||||
}
|
||||
|
||||
/*
|
||||
* Class: org_dmlc_xgboost4j_XgboostJNI
|
||||
* Method: XGBoosterCreate
|
||||
* Signature: ([J)J
|
||||
*/
|
||||
JNIEXPORT jint JNICALL Java_org_dmlc_xgboost4j_XgboostJNI_XGBoosterCreate
|
||||
(JNIEnv *jenv, jclass jcls, jlongArray jhandles, jlongArray jout) {
|
||||
DMatrixHandle* handles;
|
||||
bst_ulong len = 0;
|
||||
jlong* cjhandles = 0;
|
||||
BoosterHandle result;
|
||||
|
||||
if(jhandles) {
|
||||
len = (bst_ulong)jenv->GetArrayLength(jhandles);
|
||||
handles = new DMatrixHandle[len];
|
||||
//put handle from jhandles to chandles
|
||||
cjhandles = jenv->GetLongArrayElements(jhandles, 0);
|
||||
for(bst_ulong i=0; i<len; i++) {
|
||||
handles[i] = (DMatrixHandle) cjhandles[i];
|
||||
}
|
||||
}
|
||||
|
||||
int ret = XGBoosterCreate(handles, len, &result);
|
||||
//release
|
||||
if(jhandles) {
|
||||
delete[] handles;
|
||||
jenv->ReleaseLongArrayElements(jhandles, cjhandles, 0);
|
||||
}
|
||||
setHandle(jenv, jout, result);
|
||||
|
||||
return ret;
|
||||
}
|
||||
|
||||
/*
|
||||
* Class: org_dmlc_xgboost4j_XgboostJNI
|
||||
* Method: XGBoosterFree
|
||||
* Signature: (J)V
|
||||
*/
|
||||
JNIEXPORT jint JNICALL Java_org_dmlc_xgboost4j_XgboostJNI_XGBoosterFree
|
||||
(JNIEnv *jenv, jclass jcls, jlong jhandle) {
|
||||
BoosterHandle handle = (BoosterHandle) jhandle;
|
||||
return XGBoosterFree(handle);
|
||||
}
|
||||
|
||||
|
||||
/*
|
||||
* Class: org_dmlc_xgboost4j_XgboostJNI
|
||||
* Method: XGBoosterSetParam
|
||||
* Signature: (JLjava/lang/String;Ljava/lang/String;)V
|
||||
*/
|
||||
JNIEXPORT jint JNICALL Java_org_dmlc_xgboost4j_XgboostJNI_XGBoosterSetParam
|
||||
(JNIEnv *jenv, jclass jcls, jlong jhandle, jstring jname, jstring jvalue) {
|
||||
BoosterHandle handle = (BoosterHandle) jhandle;
|
||||
const char* name = jenv->GetStringUTFChars(jname, 0);
|
||||
const char* value = jenv->GetStringUTFChars(jvalue, 0);
|
||||
int ret = XGBoosterSetParam(handle, name, value);
|
||||
//release
|
||||
if (name) jenv->ReleaseStringUTFChars(jname, name);
|
||||
if (value) jenv->ReleaseStringUTFChars(jvalue, value);
|
||||
return ret;
|
||||
}
|
||||
|
||||
/*
|
||||
* Class: org_dmlc_xgboost4j_XgboostJNI
|
||||
* Method: XGBoosterUpdateOneIter
|
||||
* Signature: (JIJ)V
|
||||
*/
|
||||
JNIEXPORT jint JNICALL Java_org_dmlc_xgboost4j_XgboostJNI_XGBoosterUpdateOneIter
|
||||
(JNIEnv *jenv, jclass jcls, jlong jhandle, jint jiter, jlong jdtrain) {
|
||||
BoosterHandle handle = (BoosterHandle) jhandle;
|
||||
DMatrixHandle dtrain = (DMatrixHandle) jdtrain;
|
||||
return XGBoosterUpdateOneIter(handle, jiter, dtrain);
|
||||
}
|
||||
|
||||
/*
|
||||
* Class: org_dmlc_xgboost4j_XgboostJNI
|
||||
* Method: XGBoosterBoostOneIter
|
||||
* Signature: (JJ[F[F)V
|
||||
*/
|
||||
JNIEXPORT jint JNICALL Java_org_dmlc_xgboost4j_XgboostJNI_XGBoosterBoostOneIter
|
||||
(JNIEnv *jenv, jclass jcls, jlong jhandle, jlong jdtrain, jfloatArray jgrad, jfloatArray jhess) {
|
||||
BoosterHandle handle = (BoosterHandle) jhandle;
|
||||
DMatrixHandle dtrain = (DMatrixHandle) jdtrain;
|
||||
jfloat* grad = jenv->GetFloatArrayElements(jgrad, 0);
|
||||
jfloat* hess = jenv->GetFloatArrayElements(jhess, 0);
|
||||
bst_ulong len = (bst_ulong)jenv->GetArrayLength(jgrad);
|
||||
int ret = XGBoosterBoostOneIter(handle, dtrain, grad, hess, len);
|
||||
//release
|
||||
jenv->ReleaseFloatArrayElements(jgrad, grad, 0);
|
||||
jenv->ReleaseFloatArrayElements(jhess, hess, 0);
|
||||
return ret;
|
||||
}
|
||||
|
||||
/*
|
||||
* Class: org_dmlc_xgboost4j_XgboostJNI
|
||||
* Method: XGBoosterEvalOneIter
|
||||
* Signature: (JI[J[Ljava/lang/String;)Ljava/lang/String;
|
||||
*/
|
||||
JNIEXPORT jint JNICALL Java_org_dmlc_xgboost4j_XgboostJNI_XGBoosterEvalOneIter
|
||||
(JNIEnv *jenv, jclass jcls, jlong jhandle, jint jiter, jlongArray jdmats, jobjectArray jevnames, jobjectArray jout) {
|
||||
BoosterHandle handle = (BoosterHandle) jhandle;
|
||||
DMatrixHandle* dmats = 0;
|
||||
char **evnames = 0;
|
||||
char *result = 0;
|
||||
bst_ulong len = (bst_ulong)jenv->GetArrayLength(jdmats);
|
||||
if(len > 0) {
|
||||
dmats = new DMatrixHandle[len];
|
||||
evnames = new char*[len];
|
||||
}
|
||||
//put handle from jhandles to chandles
|
||||
jlong* cjdmats = jenv->GetLongArrayElements(jdmats, 0);
|
||||
for(bst_ulong i=0; i<len; i++) {
|
||||
dmats[i] = (DMatrixHandle) cjdmats[i];
|
||||
}
|
||||
//transfer jObjectArray to char**, user strcpy and release JNI char* inplace
|
||||
for(bst_ulong i=0; i<len; i++) {
|
||||
jstring jevname = (jstring)jenv->GetObjectArrayElement(jevnames, i);
|
||||
const char* cevname = jenv->GetStringUTFChars(jevname, 0);
|
||||
evnames[i] = new char[jenv->GetStringLength(jevname)];
|
||||
strcpy(evnames[i], cevname);
|
||||
jenv->ReleaseStringUTFChars(jevname, cevname);
|
||||
}
|
||||
|
||||
int ret = XGBoosterEvalOneIter(handle, jiter, dmats, (char const *(*)) evnames, len, (const char **) &result);
|
||||
if(len > 0) {
|
||||
delete[] dmats;
|
||||
//release string chars
|
||||
for(bst_ulong i=0; i<len; i++) {
|
||||
delete[] evnames[i];
|
||||
}
|
||||
delete[] evnames;
|
||||
jenv->ReleaseLongArrayElements(jdmats, cjdmats, 0);
|
||||
}
|
||||
|
||||
jstring jinfo = 0;
|
||||
if (result) jinfo = jenv->NewStringUTF((const char *) result);
|
||||
jenv->SetObjectArrayElement(jout, 0, jinfo);
|
||||
|
||||
return ret;
|
||||
}
|
||||
|
||||
/*
|
||||
* Class: org_dmlc_xgboost4j_XgboostJNI
|
||||
* Method: XGBoosterPredict
|
||||
* Signature: (JJIJ)[F
|
||||
*/
|
||||
JNIEXPORT jint JNICALL Java_org_dmlc_xgboost4j_XgboostJNI_XGBoosterPredict
|
||||
(JNIEnv *jenv, jclass jcls, jlong jhandle, jlong jdmat, jint joption_mask, jint jntree_limit, jobjectArray jout) {
|
||||
BoosterHandle handle = (BoosterHandle) jhandle;
|
||||
DMatrixHandle dmat = (DMatrixHandle) jdmat;
|
||||
bst_ulong len;
|
||||
float *result;
|
||||
int ret = XGBoosterPredict(handle, dmat, joption_mask, (unsigned int) jntree_limit, &len, (const float **) &result);
|
||||
|
||||
jsize jlen = (jsize) len;
|
||||
jfloatArray jarray = jenv->NewFloatArray(jlen);
|
||||
jenv->SetFloatArrayRegion(jarray, 0, jlen, (jfloat *) result);
|
||||
jenv->SetObjectArrayElement(jout, 0, jarray);
|
||||
return ret;
|
||||
}
|
||||
|
||||
/*
|
||||
* Class: org_dmlc_xgboost4j_XgboostJNI
|
||||
* Method: XGBoosterLoadModel
|
||||
* Signature: (JLjava/lang/String;)V
|
||||
*/
|
||||
JNIEXPORT jint JNICALL Java_org_dmlc_xgboost4j_XgboostJNI_XGBoosterLoadModel
|
||||
(JNIEnv *jenv, jclass jcls, jlong jhandle, jstring jfname) {
|
||||
BoosterHandle handle = (BoosterHandle) jhandle;
|
||||
const char* fname = jenv->GetStringUTFChars(jfname, 0);
|
||||
|
||||
int ret = XGBoosterLoadModel(handle, fname);
|
||||
if (fname) jenv->ReleaseStringUTFChars(jfname,fname);
|
||||
return ret;
|
||||
}
|
||||
|
||||
/*
|
||||
* Class: org_dmlc_xgboost4j_XgboostJNI
|
||||
* Method: XGBoosterSaveModel
|
||||
* Signature: (JLjava/lang/String;)V
|
||||
*/
|
||||
JNIEXPORT jint JNICALL Java_org_dmlc_xgboost4j_XgboostJNI_XGBoosterSaveModel
|
||||
(JNIEnv *jenv, jclass jcls, jlong jhandle, jstring jfname) {
|
||||
BoosterHandle handle = (BoosterHandle) jhandle;
|
||||
const char* fname = jenv->GetStringUTFChars(jfname, 0);
|
||||
|
||||
int ret = XGBoosterSaveModel(handle, fname);
|
||||
if (fname) jenv->ReleaseStringUTFChars(jfname, fname);
|
||||
|
||||
return ret;
|
||||
}
|
||||
|
||||
/*
|
||||
* Class: org_dmlc_xgboost4j_XgboostJNI
|
||||
* Method: XGBoosterLoadModelFromBuffer
|
||||
* Signature: (JJJ)V
|
||||
*/
|
||||
JNIEXPORT jint JNICALL Java_org_dmlc_xgboost4j_XgboostJNI_XGBoosterLoadModelFromBuffer
|
||||
(JNIEnv *jenv, jclass jcls, jlong jhandle, jlong jbuf, jlong jlen) {
|
||||
BoosterHandle handle = (BoosterHandle) jhandle;
|
||||
void *buf = (void*) jbuf;
|
||||
return XGBoosterLoadModelFromBuffer(handle, (void const *)buf, (bst_ulong) jlen);
|
||||
}
|
||||
|
||||
/*
|
||||
* Class: org_dmlc_xgboost4j_XgboostJNI
|
||||
* Method: XGBoosterGetModelRaw
|
||||
* Signature: (J)Ljava/lang/String;
|
||||
*/
|
||||
JNIEXPORT jint JNICALL Java_org_dmlc_xgboost4j_XgboostJNI_XGBoosterGetModelRaw
|
||||
(JNIEnv * jenv, jclass jcls, jlong jhandle, jobjectArray jout) {
|
||||
BoosterHandle handle = (BoosterHandle) jhandle;
|
||||
bst_ulong len = 0;
|
||||
char *result;
|
||||
|
||||
int ret = XGBoosterGetModelRaw(handle, &len, (const char **) &result);
|
||||
if (result){
|
||||
jstring jinfo = jenv->NewStringUTF((const char *) result);
|
||||
jenv->SetObjectArrayElement(jout, 0, jinfo);
|
||||
}
|
||||
return ret;
|
||||
}
|
||||
|
||||
/*
|
||||
* Class: org_dmlc_xgboost4j_XgboostJNI
|
||||
* Method: XGBoosterDumpModel
|
||||
* Signature: (JLjava/lang/String;I)[Ljava/lang/String;
|
||||
*/
|
||||
JNIEXPORT jint JNICALL Java_org_dmlc_xgboost4j_XgboostJNI_XGBoosterDumpModel
|
||||
(JNIEnv *jenv, jclass jcls, jlong jhandle, jstring jfmap, jint jwith_stats, jobjectArray jout) {
|
||||
BoosterHandle handle = (BoosterHandle) jhandle;
|
||||
const char *fmap = jenv->GetStringUTFChars(jfmap, 0);
|
||||
bst_ulong len = 0;
|
||||
char **result;
|
||||
|
||||
int ret = XGBoosterDumpModel(handle, fmap, jwith_stats, &len, (const char ***) &result);
|
||||
|
||||
jsize jlen = (jsize) len;
|
||||
jobjectArray jinfos = jenv->NewObjectArray(jlen, jenv->FindClass("java/lang/String"), jenv->NewStringUTF(""));
|
||||
for(int i=0 ; i<jlen; i++) {
|
||||
jenv->SetObjectArrayElement(jinfos, i, jenv->NewStringUTF((const char*) result[i]));
|
||||
}
|
||||
jenv->SetObjectArrayElement(jout, 0, jinfos);
|
||||
|
||||
if (fmap) jenv->ReleaseStringUTFChars(jfmap, (const char *)fmap);
|
||||
return ret;
|
||||
}
|
||||
221
jvm-packages/xgboost4j/src/native/xgboost4j.h
Normal file
221
jvm-packages/xgboost4j/src/native/xgboost4j.h
Normal file
@@ -0,0 +1,221 @@
|
||||
/* DO NOT EDIT THIS FILE - it is machine generated */
|
||||
#include <jni.h>
|
||||
/* Header for class org_dmlc_xgboost4j_XgboostJNI */
|
||||
|
||||
#ifndef _Included_org_dmlc_xgboost4j_XgboostJNI
|
||||
#define _Included_org_dmlc_xgboost4j_XgboostJNI
|
||||
#ifdef __cplusplus
|
||||
extern "C" {
|
||||
#endif
|
||||
/*
|
||||
* Class: org_dmlc_xgboost4j_XgboostJNI
|
||||
* Method: XGBGetLastError
|
||||
* Signature: ()Ljava/lang/String;
|
||||
*/
|
||||
JNIEXPORT jstring JNICALL Java_org_dmlc_xgboost4j_XgboostJNI_XGBGetLastError
|
||||
(JNIEnv *, jclass);
|
||||
|
||||
/*
|
||||
* Class: org_dmlc_xgboost4j_XgboostJNI
|
||||
* Method: XGDMatrixCreateFromFile
|
||||
* Signature: (Ljava/lang/String;I[J)I
|
||||
*/
|
||||
JNIEXPORT jint JNICALL Java_org_dmlc_xgboost4j_XgboostJNI_XGDMatrixCreateFromFile
|
||||
(JNIEnv *, jclass, jstring, jint, jlongArray);
|
||||
|
||||
/*
|
||||
* Class: org_dmlc_xgboost4j_XgboostJNI
|
||||
* Method: XGDMatrixCreateFromCSR
|
||||
* Signature: ([J[I[F[J)I
|
||||
*/
|
||||
JNIEXPORT jint JNICALL Java_org_dmlc_xgboost4j_XgboostJNI_XGDMatrixCreateFromCSR
|
||||
(JNIEnv *, jclass, jlongArray, jintArray, jfloatArray, jlongArray);
|
||||
|
||||
/*
|
||||
* Class: org_dmlc_xgboost4j_XgboostJNI
|
||||
* Method: XGDMatrixCreateFromCSC
|
||||
* Signature: ([J[I[F[J)I
|
||||
*/
|
||||
JNIEXPORT jint JNICALL Java_org_dmlc_xgboost4j_XgboostJNI_XGDMatrixCreateFromCSC
|
||||
(JNIEnv *, jclass, jlongArray, jintArray, jfloatArray, jlongArray);
|
||||
|
||||
/*
|
||||
* Class: org_dmlc_xgboost4j_XgboostJNI
|
||||
* Method: XGDMatrixCreateFromMat
|
||||
* Signature: ([FIIF[J)I
|
||||
*/
|
||||
JNIEXPORT jint JNICALL Java_org_dmlc_xgboost4j_XgboostJNI_XGDMatrixCreateFromMat
|
||||
(JNIEnv *, jclass, jfloatArray, jint, jint, jfloat, jlongArray);
|
||||
|
||||
/*
|
||||
* Class: org_dmlc_xgboost4j_XgboostJNI
|
||||
* Method: XGDMatrixSliceDMatrix
|
||||
* Signature: (J[I[J)I
|
||||
*/
|
||||
JNIEXPORT jint JNICALL Java_org_dmlc_xgboost4j_XgboostJNI_XGDMatrixSliceDMatrix
|
||||
(JNIEnv *, jclass, jlong, jintArray, jlongArray);
|
||||
|
||||
/*
|
||||
* Class: org_dmlc_xgboost4j_XgboostJNI
|
||||
* Method: XGDMatrixFree
|
||||
* Signature: (J)I
|
||||
*/
|
||||
JNIEXPORT jint JNICALL Java_org_dmlc_xgboost4j_XgboostJNI_XGDMatrixFree
|
||||
(JNIEnv *, jclass, jlong);
|
||||
|
||||
/*
|
||||
* Class: org_dmlc_xgboost4j_XgboostJNI
|
||||
* Method: XGDMatrixSaveBinary
|
||||
* Signature: (JLjava/lang/String;I)I
|
||||
*/
|
||||
JNIEXPORT jint JNICALL Java_org_dmlc_xgboost4j_XgboostJNI_XGDMatrixSaveBinary
|
||||
(JNIEnv *, jclass, jlong, jstring, jint);
|
||||
|
||||
/*
|
||||
* Class: org_dmlc_xgboost4j_XgboostJNI
|
||||
* Method: XGDMatrixSetFloatInfo
|
||||
* Signature: (JLjava/lang/String;[F)I
|
||||
*/
|
||||
JNIEXPORT jint JNICALL Java_org_dmlc_xgboost4j_XgboostJNI_XGDMatrixSetFloatInfo
|
||||
(JNIEnv *, jclass, jlong, jstring, jfloatArray);
|
||||
|
||||
/*
|
||||
* Class: org_dmlc_xgboost4j_XgboostJNI
|
||||
* Method: XGDMatrixSetUIntInfo
|
||||
* Signature: (JLjava/lang/String;[I)I
|
||||
*/
|
||||
JNIEXPORT jint JNICALL Java_org_dmlc_xgboost4j_XgboostJNI_XGDMatrixSetUIntInfo
|
||||
(JNIEnv *, jclass, jlong, jstring, jintArray);
|
||||
|
||||
/*
|
||||
* Class: org_dmlc_xgboost4j_XgboostJNI
|
||||
* Method: XGDMatrixSetGroup
|
||||
* Signature: (J[I)I
|
||||
*/
|
||||
JNIEXPORT jint JNICALL Java_org_dmlc_xgboost4j_XgboostJNI_XGDMatrixSetGroup
|
||||
(JNIEnv *, jclass, jlong, jintArray);
|
||||
|
||||
/*
|
||||
* Class: org_dmlc_xgboost4j_XgboostJNI
|
||||
* Method: XGDMatrixGetFloatInfo
|
||||
* Signature: (JLjava/lang/String;[[F)I
|
||||
*/
|
||||
JNIEXPORT jint JNICALL Java_org_dmlc_xgboost4j_XgboostJNI_XGDMatrixGetFloatInfo
|
||||
(JNIEnv *, jclass, jlong, jstring, jobjectArray);
|
||||
|
||||
/*
|
||||
* Class: org_dmlc_xgboost4j_XgboostJNI
|
||||
* Method: XGDMatrixGetUIntInfo
|
||||
* Signature: (JLjava/lang/String;[[I)I
|
||||
*/
|
||||
JNIEXPORT jint JNICALL Java_org_dmlc_xgboost4j_XgboostJNI_XGDMatrixGetUIntInfo
|
||||
(JNIEnv *, jclass, jlong, jstring, jobjectArray);
|
||||
|
||||
/*
|
||||
* Class: org_dmlc_xgboost4j_XgboostJNI
|
||||
* Method: XGDMatrixNumRow
|
||||
* Signature: (J[J)I
|
||||
*/
|
||||
JNIEXPORT jint JNICALL Java_org_dmlc_xgboost4j_XgboostJNI_XGDMatrixNumRow
|
||||
(JNIEnv *, jclass, jlong, jlongArray);
|
||||
|
||||
/*
|
||||
* Class: org_dmlc_xgboost4j_XgboostJNI
|
||||
* Method: XGBoosterCreate
|
||||
* Signature: ([J[J)I
|
||||
*/
|
||||
JNIEXPORT jint JNICALL Java_org_dmlc_xgboost4j_XgboostJNI_XGBoosterCreate
|
||||
(JNIEnv *, jclass, jlongArray, jlongArray);
|
||||
|
||||
/*
|
||||
* Class: org_dmlc_xgboost4j_XgboostJNI
|
||||
* Method: XGBoosterFree
|
||||
* Signature: (J)I
|
||||
*/
|
||||
JNIEXPORT jint JNICALL Java_org_dmlc_xgboost4j_XgboostJNI_XGBoosterFree
|
||||
(JNIEnv *, jclass, jlong);
|
||||
|
||||
/*
|
||||
* Class: org_dmlc_xgboost4j_XgboostJNI
|
||||
* Method: XGBoosterSetParam
|
||||
* Signature: (JLjava/lang/String;Ljava/lang/String;)I
|
||||
*/
|
||||
JNIEXPORT jint JNICALL Java_org_dmlc_xgboost4j_XgboostJNI_XGBoosterSetParam
|
||||
(JNIEnv *, jclass, jlong, jstring, jstring);
|
||||
|
||||
/*
|
||||
* Class: org_dmlc_xgboost4j_XgboostJNI
|
||||
* Method: XGBoosterUpdateOneIter
|
||||
* Signature: (JIJ)I
|
||||
*/
|
||||
JNIEXPORT jint JNICALL Java_org_dmlc_xgboost4j_XgboostJNI_XGBoosterUpdateOneIter
|
||||
(JNIEnv *, jclass, jlong, jint, jlong);
|
||||
|
||||
/*
|
||||
* Class: org_dmlc_xgboost4j_XgboostJNI
|
||||
* Method: XGBoosterBoostOneIter
|
||||
* Signature: (JJ[F[F)I
|
||||
*/
|
||||
JNIEXPORT jint JNICALL Java_org_dmlc_xgboost4j_XgboostJNI_XGBoosterBoostOneIter
|
||||
(JNIEnv *, jclass, jlong, jlong, jfloatArray, jfloatArray);
|
||||
|
||||
/*
|
||||
* Class: org_dmlc_xgboost4j_XgboostJNI
|
||||
* Method: XGBoosterEvalOneIter
|
||||
* Signature: (JI[J[Ljava/lang/String;[Ljava/lang/String;)I
|
||||
*/
|
||||
JNIEXPORT jint JNICALL Java_org_dmlc_xgboost4j_XgboostJNI_XGBoosterEvalOneIter
|
||||
(JNIEnv *, jclass, jlong, jint, jlongArray, jobjectArray, jobjectArray);
|
||||
|
||||
/*
|
||||
* Class: org_dmlc_xgboost4j_XgboostJNI
|
||||
* Method: XGBoosterPredict
|
||||
* Signature: (JJIJ[[F)I
|
||||
*/
|
||||
JNIEXPORT jint JNICALL Java_org_dmlc_xgboost4j_XgboostJNI_XGBoosterPredict
|
||||
(JNIEnv *, jclass, jlong, jlong, jint, jint, jobjectArray);
|
||||
|
||||
/*
|
||||
* Class: org_dmlc_xgboost4j_XgboostJNI
|
||||
* Method: XGBoosterLoadModel
|
||||
* Signature: (JLjava/lang/String;)I
|
||||
*/
|
||||
JNIEXPORT jint JNICALL Java_org_dmlc_xgboost4j_XgboostJNI_XGBoosterLoadModel
|
||||
(JNIEnv *, jclass, jlong, jstring);
|
||||
|
||||
/*
|
||||
* Class: org_dmlc_xgboost4j_XgboostJNI
|
||||
* Method: XGBoosterSaveModel
|
||||
* Signature: (JLjava/lang/String;)I
|
||||
*/
|
||||
JNIEXPORT jint JNICALL Java_org_dmlc_xgboost4j_XgboostJNI_XGBoosterSaveModel
|
||||
(JNIEnv *, jclass, jlong, jstring);
|
||||
|
||||
/*
|
||||
* Class: org_dmlc_xgboost4j_XgboostJNI
|
||||
* Method: XGBoosterLoadModelFromBuffer
|
||||
* Signature: (JJJ)I
|
||||
*/
|
||||
JNIEXPORT jint JNICALL Java_org_dmlc_xgboost4j_XgboostJNI_XGBoosterLoadModelFromBuffer
|
||||
(JNIEnv *, jclass, jlong, jlong, jlong);
|
||||
|
||||
/*
|
||||
* Class: org_dmlc_xgboost4j_XgboostJNI
|
||||
* Method: XGBoosterGetModelRaw
|
||||
* Signature: (J[Ljava/lang/String;)I
|
||||
*/
|
||||
JNIEXPORT jint JNICALL Java_org_dmlc_xgboost4j_XgboostJNI_XGBoosterGetModelRaw
|
||||
(JNIEnv *, jclass, jlong, jobjectArray);
|
||||
|
||||
/*
|
||||
* Class: org_dmlc_xgboost4j_XgboostJNI
|
||||
* Method: XGBoosterDumpModel
|
||||
* Signature: (JLjava/lang/String;I[[Ljava/lang/String;)I
|
||||
*/
|
||||
JNIEXPORT jint JNICALL Java_org_dmlc_xgboost4j_XgboostJNI_XGBoosterDumpModel
|
||||
(JNIEnv *, jclass, jlong, jstring, jint, jobjectArray);
|
||||
|
||||
#ifdef __cplusplus
|
||||
}
|
||||
#endif
|
||||
#endif
|
||||
@@ -0,0 +1,139 @@
|
||||
/*
|
||||
Copyright (c) 2014 by Contributors
|
||||
|
||||
Licensed under the Apache License, Version 2.0 (the "License");
|
||||
you may not use this file except in compliance with the License.
|
||||
You may obtain a copy of the License at
|
||||
|
||||
http://www.apache.org/licenses/LICENSE-2.0
|
||||
|
||||
Unless required by applicable law or agreed to in writing, software
|
||||
distributed under the License is distributed on an "AS IS" BASIS,
|
||||
WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
|
||||
See the License for the specific language governing permissions and
|
||||
limitations under the License.
|
||||
*/
|
||||
package org.dmlc.xgboost4j;
|
||||
|
||||
import junit.framework.TestCase;
|
||||
import org.apache.commons.logging.Log;
|
||||
import org.apache.commons.logging.LogFactory;
|
||||
import org.junit.Test;
|
||||
|
||||
import java.util.*;
|
||||
import java.util.Map.Entry;
|
||||
|
||||
/**
|
||||
* test cases for Booster
|
||||
*
|
||||
* @author hzx
|
||||
*/
|
||||
public class BoosterImplTest {
|
||||
public static class EvalError implements IEvaluation {
|
||||
private static final Log logger = LogFactory.getLog(EvalError.class);
|
||||
|
||||
String evalMetric = "custom_error";
|
||||
|
||||
public EvalError() {
|
||||
}
|
||||
|
||||
@Override
|
||||
public String getMetric() {
|
||||
return evalMetric;
|
||||
}
|
||||
|
||||
@Override
|
||||
public float eval(float[][] predicts, org.dmlc.xgboost4j.DMatrix dmat) {
|
||||
float error = 0f;
|
||||
float[] labels;
|
||||
try {
|
||||
labels = dmat.getLabel();
|
||||
} catch (XGBoostError ex) {
|
||||
logger.error(ex);
|
||||
return -1f;
|
||||
}
|
||||
int nrow = predicts.length;
|
||||
for (int i = 0; i < nrow; i++) {
|
||||
if (labels[i] == 0f && predicts[i][0] > 0) {
|
||||
error++;
|
||||
} else if (labels[i] == 1f && predicts[i][0] <= 0) {
|
||||
error++;
|
||||
}
|
||||
}
|
||||
|
||||
return error / labels.length;
|
||||
}
|
||||
}
|
||||
|
||||
@Test
|
||||
public void testBoosterBasic() throws XGBoostError {
|
||||
org.dmlc.xgboost4j.DMatrix trainMat = new org.dmlc.xgboost4j.DMatrix("../../demo/data/agaricus.txt.train");
|
||||
org.dmlc.xgboost4j.DMatrix testMat = new org.dmlc.xgboost4j.DMatrix("../../demo/data/agaricus.txt.test");
|
||||
|
||||
//set params
|
||||
Map<String, Object> paramMap = new HashMap<String, Object>() {
|
||||
{
|
||||
put("eta", 1.0);
|
||||
put("max_depth", 2);
|
||||
put("silent", 1);
|
||||
put("objective", "binary:logistic");
|
||||
}
|
||||
};
|
||||
|
||||
//set watchList
|
||||
HashMap<String, org.dmlc.xgboost4j.DMatrix> watches = new HashMap<>();
|
||||
|
||||
watches.put("train", trainMat);
|
||||
watches.put("test", testMat);
|
||||
|
||||
//set round
|
||||
int round = 2;
|
||||
|
||||
//train a boost model
|
||||
Booster booster = XGBoost.train(paramMap, trainMat, round, watches, null, null);
|
||||
|
||||
//predict raw output
|
||||
float[][] predicts = booster.predict(testMat, true);
|
||||
|
||||
//eval
|
||||
IEvaluation eval = new EvalError();
|
||||
//error must be less than 0.1
|
||||
TestCase.assertTrue(eval.eval(predicts, testMat) < 0.1f);
|
||||
|
||||
//test dump model
|
||||
|
||||
}
|
||||
|
||||
/**
|
||||
* test cross valiation
|
||||
*
|
||||
* @throws XGBoostError
|
||||
*/
|
||||
@Test
|
||||
public void testCV() throws XGBoostError {
|
||||
//load train mat
|
||||
org.dmlc.xgboost4j.DMatrix trainMat = new org.dmlc.xgboost4j.DMatrix("../../demo/data/agaricus.txt.train");
|
||||
|
||||
//set params
|
||||
Map<String, Object> param = new HashMap<String, Object>() {
|
||||
{
|
||||
put("eta", 1.0);
|
||||
put("max_depth", 3);
|
||||
put("silent", 1);
|
||||
put("nthread", 6);
|
||||
put("objective", "binary:logistic");
|
||||
put("gamma", 1.0);
|
||||
put("eval_metric", "error");
|
||||
}
|
||||
};
|
||||
|
||||
//do 5-fold cross validation
|
||||
int round = 2;
|
||||
int nfold = 5;
|
||||
//set additional eval_metrics
|
||||
String[] metrics = null;
|
||||
|
||||
String[] evalHist = XGBoost.crossValiation(param, trainMat, round, nfold, metrics,
|
||||
null, null);
|
||||
}
|
||||
}
|
||||
@@ -0,0 +1,104 @@
|
||||
/*
|
||||
Copyright (c) 2014 by Contributors
|
||||
|
||||
Licensed under the Apache License, Version 2.0 (the "License");
|
||||
you may not use this file except in compliance with the License.
|
||||
You may obtain a copy of the License at
|
||||
|
||||
http://www.apache.org/licenses/LICENSE-2.0
|
||||
|
||||
Unless required by applicable law or agreed to in writing, software
|
||||
distributed under the License is distributed on an "AS IS" BASIS,
|
||||
WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
|
||||
See the License for the specific language governing permissions and
|
||||
limitations under the License.
|
||||
*/
|
||||
package org.dmlc.xgboost4j;
|
||||
|
||||
import junit.framework.TestCase;
|
||||
import org.dmlc.xgboost4j.*;
|
||||
import org.junit.Test;
|
||||
|
||||
import java.util.Arrays;
|
||||
import java.util.Random;
|
||||
|
||||
/**
|
||||
* test cases for DMatrix
|
||||
*
|
||||
* @author hzx
|
||||
*/
|
||||
public class DMatrixTest {
|
||||
|
||||
@Test
|
||||
public void testCreateFromFile() throws XGBoostError {
|
||||
//create DMatrix from file
|
||||
org.dmlc.xgboost4j.DMatrix dmat = new org.dmlc.xgboost4j.DMatrix("../../demo/data/agaricus.txt.test");
|
||||
//get label
|
||||
float[] labels = dmat.getLabel();
|
||||
//check length
|
||||
TestCase.assertTrue(dmat.rowNum() == labels.length);
|
||||
//set weights
|
||||
float[] weights = Arrays.copyOf(labels, labels.length);
|
||||
dmat.setWeight(weights);
|
||||
float[] dweights = dmat.getWeight();
|
||||
TestCase.assertTrue(Arrays.equals(weights, dweights));
|
||||
}
|
||||
|
||||
@Test
|
||||
public void testCreateFromCSR() throws XGBoostError {
|
||||
//create Matrix from csr format sparse Matrix and labels
|
||||
/**
|
||||
* sparse matrix
|
||||
* 1 0 2 3 0
|
||||
* 4 0 2 3 5
|
||||
* 3 1 2 5 0
|
||||
*/
|
||||
float[] data = new float[]{1, 2, 3, 4, 2, 3, 5, 3, 1, 2, 5};
|
||||
int[] colIndex = new int[]{0, 2, 3, 0, 2, 3, 4, 0, 1, 2, 3};
|
||||
long[] rowHeaders = new long[]{0, 3, 7, 11};
|
||||
org.dmlc.xgboost4j.DMatrix dmat1 = new org.dmlc.xgboost4j.DMatrix(rowHeaders, colIndex, data, org.dmlc.xgboost4j.DMatrix.SparseType.CSR);
|
||||
//check row num
|
||||
System.out.println(dmat1.rowNum());
|
||||
TestCase.assertTrue(dmat1.rowNum() == 3);
|
||||
//test set label
|
||||
float[] label1 = new float[]{1, 0, 1};
|
||||
dmat1.setLabel(label1);
|
||||
float[] label2 = dmat1.getLabel();
|
||||
TestCase.assertTrue(Arrays.equals(label1, label2));
|
||||
}
|
||||
|
||||
@Test
|
||||
public void testCreateFromDenseMatrix() throws XGBoostError {
|
||||
//create DMatrix from 10*5 dense matrix
|
||||
int nrow = 10;
|
||||
int ncol = 5;
|
||||
float[] data0 = new float[nrow * ncol];
|
||||
//put random nums
|
||||
Random random = new Random();
|
||||
for (int i = 0; i < nrow * ncol; i++) {
|
||||
data0[i] = random.nextFloat();
|
||||
}
|
||||
|
||||
//create label
|
||||
float[] label0 = new float[nrow];
|
||||
for (int i = 0; i < nrow; i++) {
|
||||
label0[i] = random.nextFloat();
|
||||
}
|
||||
|
||||
org.dmlc.xgboost4j.DMatrix dmat0 = new org.dmlc.xgboost4j.DMatrix(data0, nrow, ncol);
|
||||
dmat0.setLabel(label0);
|
||||
|
||||
//check
|
||||
TestCase.assertTrue(dmat0.rowNum() == 10);
|
||||
TestCase.assertTrue(dmat0.getLabel().length == 10);
|
||||
|
||||
//set weights for each instance
|
||||
float[] weights = new float[nrow];
|
||||
for (int i = 0; i < nrow; i++) {
|
||||
weights[i] = random.nextFloat();
|
||||
}
|
||||
dmat0.setWeight(weights);
|
||||
|
||||
TestCase.assertTrue(Arrays.equals(weights, dmat0.getWeight()));
|
||||
}
|
||||
}
|
||||
Reference in New Issue
Block a user