ai.h2o.automl

Class FrameMetadata

java.lang.Object

water.Iced

ai.h2o.automl.FrameMetadata

All Implemented Interfaces:

java.io.Externalizable, java.io.Serializable, java.lang.Cloneable, water.Freezable

public class FrameMetadata
extends water.Iced

Cache common questions asked upon the frame.

See Also:

Serialized Form

Field Summary

Fields

Modifier and Type	Field and Description
int[]	_binaryCols
int[]	_catFeats
ColMeta[]	_cols
int[]	_dblCols
water.fvec.Frame	_fr
int[]	_intCols
int[]	_intNotBinaryCols
int[]	_numFeats
int	_response
water.fvec.Vec[]	_trainTestWeight
static java.lang.String[]	METAVALUES
static double	SQLNAN

Constructor Summary

Constructors

Constructor and Description
FrameMetadata(UserFeedback userFeedback, water.fvec.Frame fr, int response, int[] predictors, java.lang.String datasetName, boolean isClassification)
FrameMetadata(UserFeedback userFeedback, water.fvec.Frame fr, int response, java.lang.String datasetName)
FrameMetadata(UserFeedback userFeedback, water.fvec.Frame fr, int response, java.lang.String[] predictors, java.lang.String datasetName, boolean isClassification)
FrameMetadata(UserFeedback userFeedback, water.fvec.Frame fr, int response, java.lang.String datasetName, boolean isClassification)

Method Summary

Methods

Modifier and Type	Method and Description
double[]	cardinality() Loops over categorical features to get cardinality summary for the frame
FrameMetadata	computeFrameMetaPass1()
void	delete()
int[]	diffCols(int[] filterThese) Get the non-ignored columns that are not in the filter; do not include the response.
static water.fvec.Vec	dvec(double... rows) A numeric Vec from an array of doubles
void	fillSimpleMeta(java.util.HashMap<java.lang.String,java.lang.Object> fm)
java.lang.String[]	ignoredCols()
java.lang.String[]	includedCols()
static java.lang.String[]	intAtoStringA(int[] select, java.lang.String[] names)
boolean	isAnyCategorical() checks if there are any categorical features in the frame
boolean	isAnyNumeric() checks if there are any numeric features in the frame
boolean	isClassification()
double[]	kurtosis() Loops over numeric features to get kurtosis summary for the frame
static java.util.HashMap<java.lang.String,java.lang.Object>	makeEmptyFrameMeta()
long	na_FeatureCount()
long	nClass()
int	numberOfCategoricalFeatures()
int	numberOfNumericFeatures() If predictors were passed, then any values computed/cached are based on those predictors
double	rapidMax(water.fvec.Frame dr) max function from rapids
double	rapidMean(water.fvec.Frame dr) mean function with default of ignore_na = true
double	rapidMean(water.fvec.Frame dr, boolean ignore_na) AstMean now accepts a flag to treat NAs as a 0 or ignore them completely
double	rapidMedian(water.fvec.Frame dr) median function from rapids
double	rapidMin(water.fvec.Frame dr) min function from rapids
double	rapidSd(water.fvec.Frame dr) sd function from rapids
ColMeta	response()
long	rowsWithNa()
double[]	skewness() Loops over numeric features to get skewness summary for the frame
boolean	stratify()
water.fvec.Vec[]	weights()

Methods inherited from class water.Iced

asBytes, clone, copyOver, frozenType, read, readExternal, readJSON, reloadFromBytes, toJsonString, write, writeExternal, writeJSON

Methods inherited from class java.lang.Object

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

Field Detail

_fr

public final water.fvec.Frame _fr

_catFeats

public int[] _catFeats

_numFeats

public int[] _numFeats

_intCols

public int[] _intCols

_dblCols

public int[] _dblCols

_binaryCols

public int[] _binaryCols

_intNotBinaryCols

public int[] _intNotBinaryCols

_response

public int _response

_cols

public ColMeta[] _cols

_trainTestWeight

public water.fvec.Vec[] _trainTestWeight

SQLNAN

public static final double SQLNAN

See Also:

Constant Field Values

METAVALUES

public static final java.lang.String[] METAVALUES

Constructor Detail

FrameMetadata

public FrameMetadata(UserFeedback userFeedback,
             water.fvec.Frame fr,
             int response,
             java.lang.String datasetName)

FrameMetadata

public FrameMetadata(UserFeedback userFeedback,
             water.fvec.Frame fr,
             int response,
             java.lang.String datasetName,
             boolean isClassification)

FrameMetadata

public FrameMetadata(UserFeedback userFeedback,
             water.fvec.Frame fr,
             int response,
             int[] predictors,
             java.lang.String datasetName,
             boolean isClassification)

FrameMetadata

public FrameMetadata(UserFeedback userFeedback,
             water.fvec.Frame fr,
             int response,
             java.lang.String[] predictors,
             java.lang.String datasetName,
             boolean isClassification)

Method Detail

delete

public void delete()

makeEmptyFrameMeta

public static java.util.HashMap<java.lang.String,java.lang.Object> makeEmptyFrameMeta()

fillSimpleMeta

public void fillSimpleMeta(java.util.HashMap<java.lang.String,java.lang.Object> fm)

diffCols

public int[] diffCols(int[] filterThese)

Get the non-ignored columns that are not in the filter; do not include the response.

Parameters:

filterThese - remove these columns

Returns:

an int[] of the non-ignored column indexes

na_FeatureCount

public long na_FeatureCount()

rowsWithNa

public long rowsWithNa()

nClass

public long nClass()

skewness

public double[] skewness()

Loops over numeric features to get skewness summary for the frame

kurtosis

public double[] kurtosis()

Loops over numeric features to get kurtosis summary for the frame

cardinality

public double[] cardinality()

Loops over categorical features to get cardinality summary for the frame

dvec

public static water.fvec.Vec dvec(double... rows)

A numeric Vec from an array of doubles

isAnyNumeric

public boolean isAnyNumeric()

checks if there are any numeric features in the frame

isAnyCategorical

public boolean isAnyCategorical()

checks if there are any categorical features in the frame

rapidMin

public double rapidMin(water.fvec.Frame dr)

min function from rapids

rapidMax

public double rapidMax(water.fvec.Frame dr)

max function from rapids

rapidMean

public double rapidMean(water.fvec.Frame dr,
               boolean ignore_na)

AstMean now accepts a flag to treat NAs as a 0 or ignore them completely

rapidMean

public double rapidMean(water.fvec.Frame dr)

mean function with default of ignore_na = true

rapidSd

public double rapidSd(water.fvec.Frame dr)

sd function from rapids

rapidMedian

public double rapidMedian(water.fvec.Frame dr)

median function from rapids

numberOfNumericFeatures

public int numberOfNumericFeatures()

If predictors were passed, then any values computed/cached are based on those predictors

Returns:

numberOfCategoricalFeatures

public int numberOfCategoricalFeatures()

isClassification

public boolean isClassification()

ignoredCols

public java.lang.String[] ignoredCols()

includedCols

public java.lang.String[] includedCols()

response

public ColMeta response()

stratify

public boolean stratify()

weights

public water.fvec.Vec[] weights()

computeFrameMetaPass1

public FrameMetadata computeFrameMetaPass1()

intAtoStringA

public static java.lang.String[] intAtoStringA(int[] select,
                               java.lang.String[] names)