split

Documentation

Overview

Package split provides GroupBy, Agg, Permute and other functions that create and populate Splits of etable.Table data. These are powerful tools for quickly summarizing and analyzing data.

Index

• func Agg(spl *etable.Splits, colNm string, aggTyp agg.Aggs) *etable.SplitAgg
• func AggIdx(spl *etable.Splits, colIdx int, aggTyp agg.Aggs) *etable.SplitAgg
• func AggTry(spl *etable.Splits, colNm string, aggTyp agg.Aggs) (*etable.SplitAgg, error)
• func All(ix *etable.IdxView) *etable.Splits
• func Desc(spl *etable.Splits, colNm string)
• func DescIdx(spl *etable.Splits, colIdx int)
• func DescTry(spl *etable.Splits, colNm string) error
• func GroupBy(ix *etable.IdxView, colNms []string) *etable.Splits
• func GroupByFunc(ix *etable.IdxView, fun func(row int) []string) *etable.Splits
• func GroupByIdx(ix *etable.IdxView, colIdxs []int) *etable.Splits
• func GroupByTry(ix *etable.IdxView, colNms []string) (*etable.Splits, error)
• func Permuted(ix *etable.IdxView, probs []float64, names []string) (*etable.Splits, error)

Functions

func Agg

func Agg(spl *etable.Splits, colNm string, aggTyp agg.Aggs) *etable.SplitAgg

Agg performs aggregation using given standard aggregation function across all splits, and returns the SplitAgg container of the results, which are also stored in the Splits. Column is specified by name -- see Try for error msg version.

func AggIdx

func AggIdx(spl *etable.Splits, colIdx int, aggTyp agg.Aggs) *etable.SplitAgg

AggIdx performs aggregation using given standard aggregation function across all splits, and returns the SplitAgg container of the results, which are also stored in the Splits. Column is specified by index.

func AggTry

func AggTry(spl *etable.Splits, colNm string, aggTyp agg.Aggs) (*etable.SplitAgg, error)

AggTry performs aggregation using given standard aggregation function across all splits, and returns the SplitAgg container of the results, which are also stored in the Splits. Column is specified by name -- returns error for bad column name.

func All

func All(ix *etable.IdxView) *etable.Splits

All returns a single "split" with all of the rows in given view useful for leveraging the aggregation management functions in splits

func Desc

func Desc(spl *etable.Splits, colNm string)

Desc performs aggregation using standard aggregation functions across all splits, and stores results in the Splits. Column is specified by name -- see Try for error msg version.

func DescIdx

func DescIdx(spl *etable.Splits, colIdx int)

DescIdx performs aggregation using standard aggregation functions across all splits, and stores results in the Splits. Column is specified by index.

func DescTry

func DescTry(spl *etable.Splits, colNm string) error

DescTry performs aggregation using standard aggregation functions across all splits, and stores results in the Splits. Column is specified by name -- returns error for bad column name.

func GroupBy

func GroupBy(ix *etable.IdxView, colNms []string) *etable.Splits

GroupBy returns a new Splits set based on the groups of values across the given set of column names (see Try for version with error)

func GroupByFunc

func GroupByFunc(ix *etable.IdxView, fun func(row int) []string) *etable.Splits

GroupByFunc returns a new Splits set based on the given function which returns value(s) to group on for each row of the table. The function should always return the same number of values -- if it doesn't behavior is undefined.

func GroupByIdx

func GroupByIdx(ix *etable.IdxView, colIdxs []int) *etable.Splits

GroupByIdx returns a new Splits set based on the groups of values across the given set of column indexes

func GroupByTry

func GroupByTry(ix *etable.IdxView, colNms []string) (*etable.Splits, error)

GroupByTry returns a new Splits set based on the groups of values across the given set of column names. returns error for bad column names.

func Permuted

func Permuted(ix *etable.IdxView, probs []float64, names []string) (*etable.Splits, error)

Permuted generates permuted random splits of table rows, using given list of probabilities, which will be normalized to sum to 1 (error returned if sum = 0) names are optional names for each split (e.g., Train, Test) which will be used to label the Values of the resulting Splits.