org.datavec.api.transform.reduce

Interface AggregableColumnReduction

All Superinterfaces:

ColumnOp, Operation<Schema,Schema>, Serializable

All Known Implementing Classes:

GeographicMidpointReduction

public interface AggregableColumnReduction
extends Serializable, ColumnOp

Method Summary

Modifier and Type	Method and Description
List<ColumnMetaData>	getColumnOutputMetaData(List<String> newColumnName, ColumnMetaData columnInputMeta) Post-reduce: what is the metadata (type, etc) for this column? For example: a "count unique" operation on a String (StringMetaData) column would return an Integer (IntegerMetaData) column
List<String>	getColumnsOutputName(String columnInputName) Post-reduce: what is the name of the column? For example, "myColumn" -> "mean(myColumn)"
IAggregableReduceOp<Writable,List<Writable>>	reduceOp() Reduce a single column.

Methods inherited from interface org.datavec.api.transform.ColumnOp

columnName, columnNames, getInputSchema, outputColumnName, outputColumnNames, setInputSchema

Methods inherited from interface org.datavec.api.transform.Operation

transform

Method Detail

reduceOp

IAggregableReduceOp<Writable,List<Writable>> reduceOp()

Reduce a single column. Note: The List<Writable> here is a single column in a reduction window, and NOT the single row (as is usually the case for List<Writable> instances

Parameters:

columnData - The Writable objects for a column

Returns:

Writable containing the reduced data

getColumnsOutputName

List<String> getColumnsOutputName(String columnInputName)

Post-reduce: what is the name of the column? For example, "myColumn" -> "mean(myColumn)"

Parameters:

columnInputName - Name of the column before reduction

Returns:

Name of the column after the reduction

getColumnOutputMetaData

List<ColumnMetaData> getColumnOutputMetaData(List<String> newColumnName,
                                             ColumnMetaData columnInputMeta)

Post-reduce: what is the metadata (type, etc) for this column? For example: a "count unique" operation on a String (StringMetaData) column would return an Integer (IntegerMetaData) column

Parameters:

columnInputMeta - Metadata for the column, before reduce

Returns:

Metadata for the column, after the reduction