Package biz.k11i.xgboost

Class Predictor

java.lang.Object

biz.k11i.xgboost.Predictor

All Implemented Interfaces:

java.io.Serializable

public class Predictor
extends java.lang.Object
implements java.io.Serializable

Predicts using the Xgboost model.

See Also:

Serialized Form

Nested Class Summary

Nested Classes

Modifier and Type	Class	Description
(package private) static class	Predictor.ModelParam	Parameters.

Field Summary

Fields

Modifier and Type	Field	Description
private float	base_score
private GradBooster	gbm
private Predictor.ModelParam	mparam
private java.lang.String	name_gbm
private java.lang.String	name_obj
private ObjFunction	obj
private SparkModelParam	sparkModelParam

Constructor Summary

Constructors

Constructor	Description
Predictor(java.io.InputStream in)
Predictor(java.io.InputStream in, PredictorConfiguration configuration)	Instantiates with the Xgboost model

Method Summary

Modifier and Type	Method	Description
float	getBaseScore()
GradBooster	getBooster()
int	getNumClass()	Returns number of class.
java.lang.String	getObjName()
SparkModelParam	getSparkModelParam()
(package private) void	initObjFunction(PredictorConfiguration configuration)
(package private) void	initObjGbm()
private boolean	isBeforeOrEqual12()	Used e.g.
float[]	predict(FVec feat)	Generates predictions for given feature vector.
float[]	predict(FVec feat, boolean output_margin)	Generates predictions for given feature vector.
float[]	predict(FVec feat, boolean output_margin, int ntree_limit)	Generates predictions for given feature vector.
float[]	predict(FVec feat, float base_margin)	Generates predictions for given feature vector.
float[]	predict(FVec feat, float base_margin, int ntree_limit)	Generates predictions for given feature vector.
int[]	predictLeaf(FVec feat)	Predicts leaf index of each tree.
int[]	predictLeaf(FVec feat, int ntree_limit)	Predicts leaf index of each tree.
java.lang.String[]	predictLeafPath(FVec feat)	Predicts path to leaf of each tree.
java.lang.String[]	predictLeafPath(FVec feat, int ntree_limit)	Predicts path to leaf of each tree.
(package private) float[]	predictRaw(FVec feat, int ntree_limit, float base_score)
float	predictSingle(FVec feat)	Generates a prediction for given feature vector.
float	predictSingle(FVec feat, boolean output_margin)	Generates a prediction for given feature vector.
float	predictSingle(FVec feat, boolean output_margin, int ntree_limit)	Generates a prediction for given feature vector.
(package private) float	predictSingleRaw(FVec feat, int ntree_limit)
(package private) void	readParam(ModelReader reader)

Methods inherited from class java.lang.Object

clone, equals, finalize, getClass, hashCode, notify, notifyAll, toString, wait, wait, wait

Field Detail

mparam

private Predictor.ModelParam mparam

sparkModelParam

private SparkModelParam sparkModelParam

name_obj

private java.lang.String name_obj

name_gbm

private java.lang.String name_gbm

obj

private ObjFunction obj

gbm

private GradBooster gbm

base_score

private float base_score

Constructor Detail

Predictor

public Predictor(java.io.InputStream in)
          throws java.io.IOException

Throws:

java.io.IOException

Predictor

public Predictor(java.io.InputStream in,
                 PredictorConfiguration configuration)
          throws java.io.IOException

Instantiates with the Xgboost model

Parameters:

in - input stream

configuration - configuration

Throws:

java.io.IOException - If an I/O error occurs

Method Detail

readParam

void readParam(ModelReader reader)
        throws java.io.IOException

Throws:

java.io.IOException

initObjFunction

void initObjFunction(PredictorConfiguration configuration)

initObjGbm

void initObjGbm()

predict

public float[] predict(FVec feat)

Generates predictions for given feature vector.

Parameters:

feat - feature vector

Returns:

prediction values

predict

public float[] predict(FVec feat,
                       boolean output_margin)

Generates predictions for given feature vector.

Parameters:

feat - feature vector

output_margin - whether to only predict margin value instead of transformed prediction

Returns:

prediction values

predict

public float[] predict(FVec feat,
                       float base_margin)

Generates predictions for given feature vector.

Parameters:

feat - feature vector

base_margin - predict with base margin for each prediction

Returns:

prediction values

predict

public float[] predict(FVec feat,
                       float base_margin,
                       int ntree_limit)

Generates predictions for given feature vector.

Parameters:

feat - feature vector

base_margin - predict with base margin for each prediction

ntree_limit - limit the number of trees used in prediction

Returns:

prediction values

predict

public float[] predict(FVec feat,
                       boolean output_margin,
                       int ntree_limit)

Generates predictions for given feature vector.

Parameters:

feat - feature vector

output_margin - whether to only predict margin value instead of transformed prediction

ntree_limit - limit the number of trees used in prediction

Returns:

prediction values

predictRaw

float[] predictRaw(FVec feat,
                   int ntree_limit,
                   float base_score)

predictSingle

public float predictSingle(FVec feat)

Generates a prediction for given feature vector.

This method only works when the model outputs single value.

Parameters:

feat - feature vector

Returns:

prediction value

predictSingle

public float predictSingle(FVec feat,
                           boolean output_margin)

Generates a prediction for given feature vector.

This method only works when the model outputs single value.

Parameters:

feat - feature vector

output_margin - whether to only predict margin value instead of transformed prediction

Returns:

prediction value

predictSingle

public float predictSingle(FVec feat,
                           boolean output_margin,
                           int ntree_limit)

Generates a prediction for given feature vector.

This method only works when the model outputs single value.

Parameters:

feat - feature vector

output_margin - whether to only predict margin value instead of transformed prediction

ntree_limit - limit the number of trees used in prediction

Returns:

prediction value

predictSingleRaw

float predictSingleRaw(FVec feat,
                       int ntree_limit)

predictLeaf

public int[] predictLeaf(FVec feat)

Predicts leaf index of each tree.

Parameters:

feat - feature vector

Returns:

leaf indexes

predictLeaf

public int[] predictLeaf(FVec feat,
                         int ntree_limit)

Predicts leaf index of each tree.

Parameters:

feat - feature vector

ntree_limit - limit, 0 for all

Returns:

leaf indexes

predictLeafPath

public java.lang.String[] predictLeafPath(FVec feat)

Predicts path to leaf of each tree.

Parameters:

feat - feature vector

Returns:

leaf paths

predictLeafPath

public java.lang.String[] predictLeafPath(FVec feat,
                                          int ntree_limit)

Predicts path to leaf of each tree.

Parameters:

feat - feature vector

ntree_limit - limit, 0 for all

Returns:

leaf paths

getSparkModelParam

public SparkModelParam getSparkModelParam()

getNumClass

public int getNumClass()

Returns number of class.

Returns:

number of class

isBeforeOrEqual12

private boolean isBeforeOrEqual12()

Used e.g. for the change od floating point operation order in between xgboost 1.2 and 1.3

Returns:

True if the booster was build with xgboost version <= 1.2.

getBooster

public GradBooster getBooster()

getObjName

public java.lang.String getObjName()

getBaseScore

public float getBaseScore()