Constructor and Description |
---|
DTree.Split(int col,
int bin,
boolean equal,
double se0,
double se1,
long n0,
long n1,
double p0,
double p1) |
Modifier and Type | Method and Description |
---|---|
static java.lang.StringBuilder |
ary2str(java.lang.StringBuilder sb,
int w,
double[] xs) |
static java.lang.StringBuilder |
ary2str(java.lang.StringBuilder sb,
int w,
float[] xs) |
static java.lang.StringBuilder |
ary2str(java.lang.StringBuilder sb,
int w,
long[] xs) |
int |
bin() |
int |
col() |
float |
improvement()
Returns empirical improvement in mean-squared error.
|
long |
rowsLeft() |
long |
rowsRight() |
double |
se() |
DHistogram[] |
split(int way,
char nbins,
int min_rows,
DHistogram[] hs,
float splat) |
java.lang.String |
toString() |
clone, frozenType, init, newInstance, read, toDocField, write, writeJSON, writeJSONFields
public DTree.Split(int col, int bin, boolean equal, double se0, double se1, long n0, long n1, double p0, double p1)
public final double se()
public final int col()
public final int bin()
public final long rowsLeft()
public final long rowsRight()
public final float improvement()
Formula for node splitting space into two subregions R1,R2 with predictions y1, y2:
i2(R1,R2) ~ w1*w2 / (w1+w2) * (y1 - y2)^2
For more information see (35), (45) in the paper J. Friedman - Greedy Function Approximation: A Gradient boosting machine
public DHistogram[] split(int way, char nbins, int min_rows, DHistogram[] hs, float splat)
public static java.lang.StringBuilder ary2str(java.lang.StringBuilder sb, int w, long[] xs)
public static java.lang.StringBuilder ary2str(java.lang.StringBuilder sb, int w, float[] xs)
public static java.lang.StringBuilder ary2str(java.lang.StringBuilder sb, int w, double[] xs)
public java.lang.String toString()
toString
in class java.lang.Object