Package com.cupoftea.table.interfaces

Class IDataFrame

java.lang.Object

com.cupoftea.table.interfaces.IDataFrame

Direct Known Subclasses:

DataFrame

public abstract class IDataFrame
extends Object

Field Summary

Fields

Modifier and Type	Field	Description
protected ExecutorService	executor
protected Header	header
protected Index	index
protected String	indexColumn
protected boolean	parallel
protected int	parallelThreshold
protected ArrayList<HashMap<String,Object>>	rows

Constructor Summary

Constructors

Constructor	Description
IDataFrame()
IDataFrame(Schema schema)

Method Summary

Modifier and Type	Method	Description
abstract IDataFrame	addColumn(int index, String columnName)
abstract IDataFrame	addColumn(int index, String columnName, @Nullable Object data)
abstract IDataFrame	addColumn(int index, String columnName, @Nullable ArrayList<Object> data)
abstract IDataFrame	addColumn(int index, String columnName, IColumnCreator columnCreator)
abstract IDataFrame	addColumn(int index, String columnName, IColumnCreator columnCreator, Class<?> columnType)
abstract IDataFrame	addColumn(int index, String columnName, Class<?> columnType)
abstract IDataFrame	addColumn(int index, String columnName, Class<?> columnType, IColumnCreator columnCreator)
abstract IDataFrame	addColumn(String columnName)
abstract IDataFrame	addColumn(String columnName, IColumnCreator columnCreator)
abstract IDataFrame	addColumn(String columnName, IColumnCreator columnCreator, Class<?> columnType)
abstract IDataFrame	addColumn(String columnName, Class<?> columnType)
abstract IDataFrame	addColumn(String columnName, Class<?> columnType, IColumnCreator columnCreator)
abstract IDataFrame	addRow(@NotNull Map<String,Object> row)
IDataFrame	agg(Aggregate... aggregates)
abstract IDataFrame	aggregate(String column, String separator)
abstract IDataFrame	alterColumnType(String columnName, Class<?> newColumnType)
IDataFrame	asOfJoin(IDataFrame right, String on)
IDataFrame	asOfJoin(IDataFrame right, String on, AsOfDirection direction)
IDataFrame	asOfJoin(IDataFrame right, String on, String by)
IDataFrame	asOfJoin(IDataFrame right, String on, String by, AsOfDirection direction)
IDataFrame	asOfJoin(IDataFrame right, String leftOn, String rightOn, String leftBy, String rightBy, AsOfDirection direction)
IDataFrame	asOfJoin(IDataFrame right, String leftOn, String rightOn, List<String> leftBy, List<String> rightBy, AsOfDirection direction, String leftSuffix, String rightSuffix)
IDataFrame	asOfJoin(IDataFrame right, String on, List<String> by, AsOfDirection direction)
IDataFrame	asParallel()	Alias for parallel().
IDataFrame	asParallel(@NotNull ExecutorService executor)	Alias for parallel(ExecutorService).
IDataFrame	asSequential()	Alias for sequential().
abstract IDataFrame	average(String column)
Double	averageOf(String column)
IDataFrame	categorize(String... columnNames)
void	clear()
void	clearRows()
abstract ColumnRef	col(String columnName)
static IDataFrame	concat(int axis, IDataFrame... frames)
static IDataFrame	concat(ConcatAxis axis, IDataFrame... frames)
IDataFrame	concat(IDataFrame other)	Concatenates this DataFrame with another DataFrame along rows (vertical concatenation) using outer join semantics.
static IDataFrame	concat(IDataFrame... frames)
IDataFrame	concat(IDataFrame other, boolean ignoreIndex)	Concatenates this DataFrame with another DataFrame along rows with optional index resetting.
IDataFrame	concat(IDataFrame other, int axis)	Concatenates this DataFrame with another DataFrame along the specified axis (0 for ROWS, 1 for COLUMNS) using outer join semantics.
IDataFrame	concat(IDataFrame other, int axis, JoinType how)	Concatenates this DataFrame with another DataFrame along the specified axis using the given join type.
IDataFrame	concat(IDataFrame other, int axis, String how)	Concatenates this DataFrame with another DataFrame along the specified axis using the given join type name.
IDataFrame	concat(IDataFrame other, ConcatAxis axis)	Concatenates this DataFrame with another DataFrame along the specified axis using outer join semantics.
IDataFrame	concat(IDataFrame other, ConcatAxis axis, JoinType how)	Concatenates this DataFrame with another DataFrame along the specified axis using the given join type.
IDataFrame	concat(IDataFrame other, ConcatAxis axis, JoinType how, boolean ignoreIndex)	Concatenates this DataFrame with another DataFrame with full control over axis, join type, and index resetting.
IDataFrame	concat(IDataFrame other, ConcatAxis axis, String how)	Concatenates this DataFrame with another DataFrame along the specified axis using the given join type name.
IDataFrame	concat(IDataFrame other, JoinType how)	Concatenates this DataFrame with another DataFrame along rows using the specified join type.
IDataFrame	concat(IDataFrame other, String how)	Concatenates this DataFrame with another DataFrame along rows using the specified join type name ("outer" or "inner").
static IDataFrame	concat(List<? extends IDataFrame> frames)
static IDataFrame	concat(List<? extends IDataFrame> frames, int axis)
static IDataFrame	concat(List<? extends IDataFrame> frames, ConcatAxis axis)
static IDataFrame	concat(List<? extends IDataFrame> frames, ConcatAxis axis, JoinType how)
static IDataFrame	concat(List<? extends IDataFrame> frames, ConcatAxis axis, JoinType how, boolean ignoreIndex)
static IDataFrame	concat(List<? extends IDataFrame> frames, JoinType how)
IDataFrame	concatHorizontal(IDataFrame other, String leftSuffix, String rightSuffix)	Concatenates this DataFrame with another DataFrame horizontally with custom suffixes for colliding columns.
IDataFrame	concatWith(IDataFrame... others)	Concatenates this DataFrame with multiple other DataFrames along rows.
IDataFrame	concatWith(List<? extends IDataFrame> others)	Concatenates this DataFrame with a list of other DataFrames along rows.
abstract IDataFrame	copy()
protected void	copyParallelConfigTo(IDataFrame target)
abstract IDataFrame	corr()
abstract Double	corr(String col1, String col2)
abstract Double	correlation(String col1, String col2)
abstract IDataFrame	count(String column)
int	countOf(String column)
abstract Double	cov(String col1, String col2)
abstract Double	cov(String col1, String col2, boolean isSample)
abstract Double	covariance(String col1, String col2)
abstract Double	covariance(String col1, String col2, boolean isSample)
IDataFrame	crossJoin(IDataFrame other)
abstract IDataFrame	cummax(String... columns)
abstract IDataFrame	cummin(String... columns)
abstract IDataFrame	cumprod(String... columns)
abstract IDataFrame	cumsum(String... columns)
IDataFrame	decodeDictionary()
IDataFrame	decodeDictionary(String... columnNames)
IDataFrame	denseRank(String column)
abstract IDataFrame	denseRank(String column, String outputColumn)
abstract IDataFrame	describe()
IDataFrame	diff(String column)
abstract IDataFrame	diff(String... columns)
IDataFrame	diff(String column, int periods)
abstract IDataFrame	diff(String column, int periods, String outputColumn)
abstract IDataFrame	distinct()
abstract IDataFrame	distinct(Keep keep)
abstract IDataFrame	distinct(Keep keep, String... subset)
abstract IDataFrame	distinct(String... subset)
abstract IDataFrame	distinct(List<String> subset)
abstract IDataFrame	distinct(List<String> subset, Keep keep)
abstract IDataFrame	drop(String... columnNames)
abstract IDataFrame	drop(List<String> columnNames)
abstract IDataFrame	dropDuplicates()
abstract IDataFrame	dropDuplicates(Keep keep)
abstract IDataFrame	dropDuplicates(Keep keep, String... subset)
abstract IDataFrame	dropDuplicates(String... subset)
abstract IDataFrame	dropDuplicates(List<String> subset)
abstract IDataFrame	dropDuplicates(List<String> subset, Keep keep)
abstract IDataFrame	dropna()
abstract IDataFrame	dropna(String how)
abstract IDataFrame	dropna(String... columns)
abstract IDataFrame	dropna(String how, String... columns)
abstract IDataFrame	dropna(String how, List<String> columns)
abstract IDataFrame	dropna(List<String> columns)
IDataFrame	encodeDictionary()
IDataFrame	encodeDictionary(boolean useShortCodes)
IDataFrame	encodeDictionary(boolean useShortCodes, String... columnNames)
IDataFrame	encodeDictionary(String... columnNames)
ExpandingWindow	expanding()
ExpandingWindow	expanding(int minPeriods)
abstract IDataFrame	fillna(Object value)
abstract IDataFrame	fillna(String column, Object value)
abstract IDataFrame	fillna(Map<String,Object> values)
abstract IDataFrame	filter(IFilter filter)
IDataFrame	filterEquals(String columnName, Object value)
IDataFrame	filterIn(String columnName, Object... values)
IDataFrame	filterIn(String columnName, Collection<?> values)
IDataFrame	filterNotEquals(String columnName, Object value)
abstract Column	getColumnHeader(String columnName)
ColumnVector	getColumnVector(int index)
ColumnVector	getColumnVector(String columnName)
List<ColumnVector>	getColumnVectors()
ExecutorService	getExecutor()	Returns the configured ExecutorService, or ForkJoinPool.commonPool() if none was explicitly set.
abstract ArrayList<Column>	getHeaders()
String	getIndex()
Index	getIndexObject()
ArrayList<HashMap<String,Object>>	getInternalRows()	Returns row interchange maps.
int	getParallelThreshold()	Returns the minimum row count threshold below which execution stays sequential.
abstract Row	getRow(int index)	Returns the unmodifiable Row at the specified index.
List<Row>	getRows()	Returns an unmodifiable list of Row views.
abstract Schema	getSchema()
abstract IDataFrame	head()
abstract IDataFrame	head(int n)
abstract ILoc	iloc()
abstract Row	iloc(int index)
abstract Object	iloc(int rowIndex, int colIndex)
abstract Object	iloc(int rowIndex, String columnName)
IDataFrame	innerJoin(IDataFrame other, String... on)
IDataFrame	innerJoin(IDataFrame other, String leftOn, String rightOn)
IDataFrame	innerJoin(IDataFrame other, List<String> on)
IDataFrame	innerJoin(IDataFrame other, List<String> leftOn, List<String> rightOn)
boolean	isCategorical(String columnName)
boolean	isDictionaryEncoded(String columnName)
boolean	isParallel()	Returns true if parallel execution mode is enabled.
IDataFrame	join(IDataFrame other)
IDataFrame	join(IDataFrame other, JoinType how, String on)
IDataFrame	join(IDataFrame other, JoinType how, String leftOn, String rightOn)
IDataFrame	join(IDataFrame other, JoinType how, List<String> on)
IDataFrame	join(IDataFrame other, JoinType how, List<String> leftOn, List<String> rightOn)
abstract IDataFrame	join(IDataFrame other, JoinType how, List<String> leftOn, List<String> rightOn, String leftSuffix, String rightSuffix)
IDataFrame	join(IDataFrame other, String on)
IDataFrame	join(IDataFrame other, String leftOn, String rightOn)
IDataFrame	join(IDataFrame other, String how, String leftOn, String rightOn)
IDataFrame	join(IDataFrame other, String how, List<String> on)
IDataFrame	join(IDataFrame other, String how, List<String> leftOn, List<String> rightOn)
IDataFrame	join(IDataFrame other, String how, List<String> leftOn, List<String> rightOn, String leftSuffix, String rightSuffix)
IDataFrame	join(IDataFrame other, List<String> on)
IDataFrame	join(IDataFrame other, List<String> leftOn, List<String> rightOn)
abstract Double	kurt(String column)
abstract Double	kurtosis(String column)
abstract Double	kurtosis(String column, boolean isSample)
IDataFrame	lag(String column)
IDataFrame	lag(String column, int offset)
abstract IDataFrame	lag(String column, int offset, String outputColumn)
IDataFrame	lead(String column)
IDataFrame	lead(String column, int offset)
abstract IDataFrame	lead(String column, int offset, String outputColumn)
IDataFrame	leftJoin(IDataFrame other, String... on)
IDataFrame	leftJoin(IDataFrame other, String leftOn, String rightOn)
IDataFrame	leftJoin(IDataFrame other, List<String> on)
IDataFrame	leftJoin(IDataFrame other, List<String> leftOn, List<String> rightOn)
abstract Loc	loc()
abstract Row	loc(Object key)
abstract Object	loc(Object key, String column)
IDataFrame	map(String columnName, @Nullable Class<?> columnType, IColumnCreator columnCreator)
IDataFrame	map(String columnName, IColumnCreator columnCreator)
IDataFrame	map(Function<Row,? extends Map<String,Object>> rowMapper)
IDataFrame	mapColumn(String columnName, @Nullable Class<?> targetType, Function<Object,?> valueMapper)
IDataFrame	mapColumn(String columnName, Function<Object,?> valueMapper)
IDataFrame	mapDouble(String columnName, DoubleUnaryOperator mapper)
IDataFrame	mapInt(String columnName, IntUnaryOperator mapper)
IDataFrame	mapLong(String columnName, LongUnaryOperator mapper)
abstract Double	max(String column)
abstract IDataFrame	maximum(String column)
Object	maxOf(String column)
abstract Double	mean(String column)
Double	meanOf(String column)
abstract Double	median(String column)
abstract IDataFrame	melt()
abstract IDataFrame	melt(List<String> idVars)
abstract IDataFrame	melt(List<String> idVars, List<String> valueVars)
abstract IDataFrame	melt(List<String> idVars, List<String> valueVars, String varName, String valueName)
abstract Double	min(String column)
abstract IDataFrame	minimum(String column)
Object	minOf(String column)
abstract IDataFrame	moveColumn(int inicialIndex, int finalIndex)
abstract IDataFrame	moveColumn(String columnName, int finalIndex)
abstract IDataFrame	orderBy(Orderer... orderer)
IDataFrame	outerJoin(IDataFrame other, String... on)
IDataFrame	outerJoin(IDataFrame other, String leftOn, String rightOn)
IDataFrame	outerJoin(IDataFrame other, List<String> on)
IDataFrame	outerJoin(IDataFrame other, List<String> leftOn, List<String> rightOn)
IDataFrame	parallel()	Enables multithreaded parallel execution mode using ForkJoinPool.commonPool() (or any previously configured executor).
IDataFrame	parallel(@NotNull ExecutorService executor)	Enables multithreaded parallel execution mode using the provided ExecutorService.
IDataFrame	pctChange(String column)
abstract IDataFrame	pctChange(String... columns)
IDataFrame	pctChange(String column, int periods)
abstract IDataFrame	pctChange(String column, int periods, String outputColumn)
IDataFrame	percentRank(String column)
abstract IDataFrame	percentRank(String column, String outputColumn)
abstract IDataFrame	pivot(String indexCol, String columnsCol, String valuesCol)
abstract IDataFrame	pivot(String indexCol, String columnsCol, String valuesCol, Aggregate agg)
abstract IDataFrame	pivot(String indexCol, String columnsCol, String valuesCol, Aggregation agg)
Plot	plot()	Obtains a declarative plotting builder for visualizing this DataFrame.
Plot	plot(String engine)	Obtains a declarative plotting builder backed by a specific plotting engine.
abstract Double	quantile(String column, double q)
IDataFrame	rank(String column)
abstract IDataFrame	rank(String column, String outputColumn)
abstract IDataFrame	removeColumn(int index)
abstract IDataFrame	removeColumn(String columnName)
abstract IDataFrame	removeRow(int index)
abstract IDataFrame	removeRows(IFilter filter)
abstract IDataFrame	renameColumn(int index, String newName)
abstract IDataFrame	renameColumn(String oldName, String newName)
abstract IDataFrame	renameColumn(HashMap<String,String> name)
Map<String,String>	render()	Jupyter / Polyglot notebook MIME representation bundle.
String	renderHtml()	Jupyter / IJava notebook MIME representation method.
abstract IDataFrame	resetIndex()
IDataFrame	rightJoin(IDataFrame other, String... on)
IDataFrame	rightJoin(IDataFrame other, String leftOn, String rightOn)
IDataFrame	rightJoin(IDataFrame other, List<String> on)
IDataFrame	rightJoin(IDataFrame other, List<String> leftOn, List<String> rightOn)
RollingWindow	rolling(int windowSize)
IDataFrame	rolling_mean(String column, int windowSize)
IDataFrame	rolling_mean(String column, int windowSize, String outputColumn)
IDataFrame	rolling_sum(String column, int windowSize)
IDataFrame	rolling_sum(String column, int windowSize, String outputColumn)
IDataFrame	rowNumber()
abstract IDataFrame	rowNumber(String outputColumn)
abstract IDataFrame	sample(double fraction)
abstract IDataFrame	sample(double fraction, boolean replace, long seed)
abstract IDataFrame	sample(double fraction, long seed)
abstract IDataFrame	sample(int n)
abstract IDataFrame	sample(int n, boolean replace)
abstract IDataFrame	sample(int n, boolean replace, long seed)
abstract IDataFrame	sample(int n, long seed)
abstract IDataFrame	select(String... columnNames)
abstract IDataFrame	select(List<String> columnNames)
IDataFrame	sequential()	Switches execution mode back to single-threaded sequential evaluation.
void	setHeaders(ArrayList<Column> headers)
abstract IDataFrame	setIndex(String columnName)
IDataFrame	setParallelThreshold(int threshold)	Sets the minimum row count threshold below which execution automatically stays sequential.
abstract IDataFrame	setSchema(Schema schema)
protected boolean	shouldExecuteParallel(int rowCount)
Integer	size()
abstract Double	skew(String column)
abstract Double	skewness(String column)
abstract Double	skewness(String column, boolean isSample)
abstract IDataFrame	slice(int fromRow, int toRow)
abstract Double	standardDeviation(String column)
abstract Double	standardDeviation(String column, boolean isSample)
abstract Double	std(String column)
abstract Double	std(String column, boolean isSample)
abstract IDataFrame	sum(String column)
Object	sumOf(String column)
abstract IDataFrame	tail()
abstract IDataFrame	tail(int n)
IDataFrame	toCategorical(String... columnNames)
String	toHtml()	Renders this DataFrame as a clean, responsive HTML table with sticky headers, data type indicators, and client-side pagination.
String	toHtml(int maxRows)	Renders this DataFrame as an HTML table displaying up to maxRows rows.
String	toHtml(TableHtmlConfiguration configuration)	Renders this DataFrame as an HTML table using the specified configuration.
abstract IDataFrame	transformColumn(String columnName, ColumnRef columnRef)
abstract IDataFrame	transformColumn(String columnName, ColumnVector vector)
abstract IDataFrame	transformColumn(String columnName, Function<ColumnVector,ColumnVector> transformer)
abstract IDataFrame	update(Map<String,Object> updateColumns)
abstract Double	var(String column)
abstract Double	var(String column, boolean isSample)
abstract Double	variance(String column)
abstract Double	variance(String column, boolean isSample)
Window	window()
abstract IDataFrame	withColumn(String columnName, ColumnRef columnRef)
abstract IDataFrame	withColumn(String columnName, ColumnVector vector)
abstract IDataFrame	withColumn(String columnName, IColumnCreator columnCreator)
abstract IDataFrame	withColumn(String columnName, Class<?> columnType, IColumnCreator columnCreator)
IDataFrame	withParallelThreshold(int threshold)	Fluent alias for setParallelThreshold(int).
DataFrameWriter	write()

Methods inherited from class java.lang.Object

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

Field Details

header

protected Header header

rows

protected ArrayList<HashMap<String,Object>> rows

indexColumn

protected String indexColumn

index

protected Index index

parallel

protected boolean parallel

executor

protected transient ExecutorService executor

parallelThreshold

protected int parallelThreshold

Constructor Details

IDataFrame

public IDataFrame()

IDataFrame

public IDataFrame(Schema schema)

Method Details

parallel

public IDataFrame parallel()

Enables multithreaded parallel execution mode using ForkJoinPool.commonPool() (or any previously configured executor).

parallel

public IDataFrame parallel(@NotNull
 @NotNull ExecutorService executor)

Enables multithreaded parallel execution mode using the provided ExecutorService.

asParallel

public IDataFrame asParallel()

Alias for parallel().

asParallel

public IDataFrame asParallel(@NotNull
 @NotNull ExecutorService executor)

Alias for parallel(ExecutorService).

sequential

public IDataFrame sequential()

Switches execution mode back to single-threaded sequential evaluation.

asSequential

public IDataFrame asSequential()

Alias for sequential().

isParallel

public boolean isParallel()

Returns true if parallel execution mode is enabled.

getExecutor

public ExecutorService getExecutor()

Returns the configured ExecutorService, or ForkJoinPool.commonPool() if none was explicitly set.

getParallelThreshold

public int getParallelThreshold()

Returns the minimum row count threshold below which execution stays sequential.

setParallelThreshold

public IDataFrame setParallelThreshold(int threshold)

Sets the minimum row count threshold below which execution automatically stays sequential.

withParallelThreshold

public IDataFrame withParallelThreshold(int threshold)

Fluent alias for setParallelThreshold(int).

shouldExecuteParallel

protected boolean shouldExecuteParallel(int rowCount)

copyParallelConfigTo

protected void copyParallelConfigTo(IDataFrame target)

getRows

public List<Row> getRows()

Returns an unmodifiable list of Row views. Evaluated lazily over the underlying columnar storage without allocating intermediate row maps.

getRow

public abstract Row getRow(int index)

Returns the unmodifiable Row at the specified index. The row columns are ordered according to the DataFrame header.

getInternalRows

public ArrayList<HashMap<String,Object>> getInternalRows()

Returns row interchange maps.

getColumnVectors

public List<ColumnVector> getColumnVectors()

getColumnVector

public ColumnVector getColumnVector(String columnName)

getColumnVector

public ColumnVector getColumnVector(int index)

getIndexObject

public Index getIndexObject()

clear

public void clear()

clearRows

public void clearRows()

copy

public abstract IDataFrame copy()

getHeaders

public abstract ArrayList<Column> getHeaders()

getColumnHeader

public abstract Column getColumnHeader(String columnName)

getSchema

public abstract Schema getSchema()

setSchema

public abstract IDataFrame setSchema(Schema schema)

moveColumn

public abstract IDataFrame moveColumn(String columnName,
 int finalIndex)

moveColumn

public abstract IDataFrame moveColumn(int inicialIndex,
 int finalIndex)

renameColumn

public abstract IDataFrame renameColumn(String oldName,
 String newName)

renameColumn

public abstract IDataFrame renameColumn(int index,
 String newName)

renameColumn

public abstract IDataFrame renameColumn(HashMap<String,String> name)

addColumn

public abstract IDataFrame addColumn(String columnName)

addColumn

public abstract IDataFrame addColumn(String columnName,
 Class<?> columnType)

addColumn

public abstract IDataFrame addColumn(int index,
 String columnName)

addColumn

public abstract IDataFrame addColumn(int index,
 String columnName,
 Class<?> columnType)

addColumn

public abstract IDataFrame addColumn(int index,
 String columnName,
 @Nullable
 @Nullable Object data)

addColumn

public abstract IDataFrame addColumn(int index,
 String columnName,
 @Nullable
 @Nullable ArrayList<Object> data)

addColumn

public abstract IDataFrame addColumn(String columnName,
 IColumnCreator columnCreator)

addColumn

public abstract IDataFrame addColumn(String columnName,
 Class<?> columnType,
 IColumnCreator columnCreator)

addColumn

public abstract IDataFrame addColumn(String columnName,
 IColumnCreator columnCreator,
 Class<?> columnType)

addColumn

public abstract IDataFrame addColumn(int index,
 String columnName,
 IColumnCreator columnCreator)

addColumn

public abstract IDataFrame addColumn(int index,
 String columnName,
 Class<?> columnType,
 IColumnCreator columnCreator)

addColumn

public abstract IDataFrame addColumn(int index,
 String columnName,
 IColumnCreator columnCreator,
 Class<?> columnType)

removeColumn

public abstract IDataFrame removeColumn(String columnName)

removeColumn

public abstract IDataFrame removeColumn(int index)

alterColumnType

public abstract IDataFrame alterColumnType(String columnName,
 Class<?> newColumnType)

col

public abstract ColumnRef col(String columnName)

withColumn

public abstract IDataFrame withColumn(String columnName,
 ColumnVector vector)

withColumn

public abstract IDataFrame withColumn(String columnName,
 ColumnRef columnRef)

withColumn

public abstract IDataFrame withColumn(String columnName,
 IColumnCreator columnCreator)

withColumn

public abstract IDataFrame withColumn(String columnName,
 Class<?> columnType,
 IColumnCreator columnCreator)

transformColumn

public abstract IDataFrame transformColumn(String columnName,
 ColumnVector vector)

transformColumn

public abstract IDataFrame transformColumn(String columnName,
 ColumnRef columnRef)

transformColumn

public abstract IDataFrame transformColumn(String columnName,
 Function<ColumnVector,ColumnVector> transformer)

addRow

public abstract IDataFrame addRow(@NotNull
 @NotNull Map<String,Object> row)

removeRow

public abstract IDataFrame removeRow(int index)

removeRows

public abstract IDataFrame removeRows(IFilter filter)

filter

public abstract IDataFrame filter(IFilter filter)

update

public abstract IDataFrame update(Map<String,Object> updateColumns)

orderBy

public abstract IDataFrame orderBy(Orderer... orderer)

encodeDictionary

public IDataFrame encodeDictionary(String... columnNames)

encodeDictionary

public IDataFrame encodeDictionary(boolean useShortCodes,
 String... columnNames)

encodeDictionary

public IDataFrame encodeDictionary()

encodeDictionary

public IDataFrame encodeDictionary(boolean useShortCodes)

decodeDictionary

public IDataFrame decodeDictionary(String... columnNames)

decodeDictionary

public IDataFrame decodeDictionary()

categorize

public IDataFrame categorize(String... columnNames)

toCategorical

public IDataFrame toCategorical(String... columnNames)

isDictionaryEncoded

public boolean isDictionaryEncoded(String columnName)

isCategorical

public boolean isCategorical(String columnName)

filterEquals

public IDataFrame filterEquals(String columnName,
 Object value)

filterNotEquals

public IDataFrame filterNotEquals(String columnName,
 Object value)

filterIn

public IDataFrame filterIn(String columnName,
 Collection<?> values)

filterIn

public IDataFrame filterIn(String columnName,
 Object... values)

map

public IDataFrame map(String columnName,
 IColumnCreator columnCreator)

map

public IDataFrame map(String columnName,
 @Nullable
 @Nullable Class<?> columnType,
 IColumnCreator columnCreator)

mapColumn

public IDataFrame mapColumn(String columnName,
 Function<Object,?> valueMapper)

mapColumn

public IDataFrame mapColumn(String columnName,
 @Nullable
 @Nullable Class<?> targetType,
 Function<Object,?> valueMapper)

mapInt

public IDataFrame mapInt(String columnName,
 IntUnaryOperator mapper)

mapLong

public IDataFrame mapLong(String columnName,
 LongUnaryOperator mapper)

mapDouble

public IDataFrame mapDouble(String columnName,
 DoubleUnaryOperator mapper)

map

public IDataFrame map(Function<Row,? extends Map<String,Object>> rowMapper)

agg

public IDataFrame agg(Aggregate... aggregates)

sumOf

public Object sumOf(String column)

meanOf

public Double meanOf(String column)

averageOf

public Double averageOf(String column)

minOf

public Object minOf(String column)

maxOf

public Object maxOf(String column)

countOf

public int countOf(String column)

select

public abstract IDataFrame select(String... columnNames)

select

public abstract IDataFrame select(List<String> columnNames)

drop

public abstract IDataFrame drop(String... columnNames)

drop

public abstract IDataFrame drop(List<String> columnNames)

head

public abstract IDataFrame head()

head

public abstract IDataFrame head(int n)

tail

public abstract IDataFrame tail()

tail

public abstract IDataFrame tail(int n)

slice

public abstract IDataFrame slice(int fromRow,
 int toRow)

iloc

public abstract Row iloc(int index)

iloc

public abstract Object iloc(int rowIndex,
 int colIndex)

iloc

public abstract Object iloc(int rowIndex,
 String columnName)

iloc

public abstract ILoc iloc()

getIndex

public String getIndex()

setIndex

public abstract IDataFrame setIndex(String columnName)

resetIndex

public abstract IDataFrame resetIndex()

loc

public abstract Row loc(Object key)

loc

public abstract Object loc(Object key,
 String column)

loc

public abstract Loc loc()

fillna

public abstract IDataFrame fillna(Object value)

fillna

public abstract IDataFrame fillna(String column,
 Object value)

fillna

public abstract IDataFrame fillna(Map<String,Object> values)

dropna

public abstract IDataFrame dropna()

dropna

public abstract IDataFrame dropna(String how)

dropna

public abstract IDataFrame dropna(String... columns)

dropna

public abstract IDataFrame dropna(List<String> columns)

dropna

public abstract IDataFrame dropna(String how,
 String... columns)

dropna

public abstract IDataFrame dropna(String how,
 List<String> columns)

sum

public abstract IDataFrame sum(String column)

average

public abstract IDataFrame average(String column)

maximum

public abstract IDataFrame maximum(String column)

minimum

public abstract IDataFrame minimum(String column)

count

public abstract IDataFrame count(String column)

aggregate

public abstract IDataFrame aggregate(String column,
 String separator)

mean

public abstract Double mean(String column)

min

public abstract Double min(String column)

max

public abstract Double max(String column)

median

public abstract Double median(String column)

variance

public abstract Double variance(String column)

variance

public abstract Double variance(String column,
 boolean isSample)

var

public abstract Double var(String column)

var

public abstract Double var(String column,
 boolean isSample)

standardDeviation

public abstract Double standardDeviation(String column)

standardDeviation

public abstract Double standardDeviation(String column,
 boolean isSample)

std

public abstract Double std(String column)

std

public abstract Double std(String column,
 boolean isSample)

quantile

public abstract Double quantile(String column,
 double q)

skewness

public abstract Double skewness(String column)

skewness

public abstract Double skewness(String column,
 boolean isSample)

skew

public abstract Double skew(String column)

kurtosis

public abstract Double kurtosis(String column)

kurtosis

public abstract Double kurtosis(String column,
 boolean isSample)

kurt

public abstract Double kurt(String column)

covariance

public abstract Double covariance(String col1,
 String col2)

covariance

public abstract Double covariance(String col1,
 String col2,
 boolean isSample)

cov

public abstract Double cov(String col1,
 String col2)

cov

public abstract Double cov(String col1,
 String col2,
 boolean isSample)

correlation

public abstract Double correlation(String col1,
 String col2)

corr

public abstract Double corr(String col1,
 String col2)

corr

public abstract IDataFrame corr()

describe

public abstract IDataFrame describe()

dropDuplicates

public abstract IDataFrame dropDuplicates()

dropDuplicates

public abstract IDataFrame dropDuplicates(Keep keep)

dropDuplicates

public abstract IDataFrame dropDuplicates(String... subset)

dropDuplicates

public abstract IDataFrame dropDuplicates(Keep keep,
 String... subset)

dropDuplicates

public abstract IDataFrame dropDuplicates(List<String> subset)

dropDuplicates

public abstract IDataFrame dropDuplicates(List<String> subset,
 Keep keep)

distinct

public abstract IDataFrame distinct()

distinct

public abstract IDataFrame distinct(Keep keep)

distinct

public abstract IDataFrame distinct(String... subset)

distinct

public abstract IDataFrame distinct(Keep keep,
 String... subset)

distinct

public abstract IDataFrame distinct(List<String> subset)

distinct

public abstract IDataFrame distinct(List<String> subset,
 Keep keep)

pivot

public abstract IDataFrame pivot(String indexCol,
 String columnsCol,
 String valuesCol)

pivot

public abstract IDataFrame pivot(String indexCol,
 String columnsCol,
 String valuesCol,
 Aggregation agg)

pivot

public abstract IDataFrame pivot(String indexCol,
 String columnsCol,
 String valuesCol,
 Aggregate agg)

melt

public abstract IDataFrame melt()

melt

public abstract IDataFrame melt(List<String> idVars)

melt

public abstract IDataFrame melt(List<String> idVars,
 List<String> valueVars)

melt

public abstract IDataFrame melt(List<String> idVars,
 List<String> valueVars,
 String varName,
 String valueName)

sample

public abstract IDataFrame sample(int n)

sample

public abstract IDataFrame sample(int n,
 boolean replace)

sample

public abstract IDataFrame sample(int n,
 long seed)

sample

public abstract IDataFrame sample(int n,
 boolean replace,
 long seed)

sample

public abstract IDataFrame sample(double fraction)

sample

public abstract IDataFrame sample(double fraction,
 long seed)

sample

public abstract IDataFrame sample(double fraction,
 boolean replace,
 long seed)

size

public Integer size()

join

public IDataFrame join(IDataFrame other)

join

public IDataFrame join(IDataFrame other,
 String on)

join

public IDataFrame join(IDataFrame other,
 List<String> on)

join

public IDataFrame join(IDataFrame other,
 JoinType how,
 String on)

join

public IDataFrame join(IDataFrame other,
 JoinType how,
 List<String> on)

join

public IDataFrame join(IDataFrame other,
 String how,
 List<String> on)

join

public IDataFrame join(IDataFrame other,
 String leftOn,
 String rightOn)

join

public IDataFrame join(IDataFrame other,
 JoinType how,
 String leftOn,
 String rightOn)

join

public IDataFrame join(IDataFrame other,
 String how,
 String leftOn,
 String rightOn)

join

public IDataFrame join(IDataFrame other,
 List<String> leftOn,
 List<String> rightOn)

join

public IDataFrame join(IDataFrame other,
 JoinType how,
 List<String> leftOn,
 List<String> rightOn)

join

public IDataFrame join(IDataFrame other,
 String how,
 List<String> leftOn,
 List<String> rightOn)

join

public IDataFrame join(IDataFrame other,
 String how,
 List<String> leftOn,
 List<String> rightOn,
 String leftSuffix,
 String rightSuffix)

join

public abstract IDataFrame join(IDataFrame other,
 JoinType how,
 List<String> leftOn,
 List<String> rightOn,
 String leftSuffix,
 String rightSuffix)

innerJoin

public IDataFrame innerJoin(IDataFrame other,
 String... on)

innerJoin

public IDataFrame innerJoin(IDataFrame other,
 List<String> on)

innerJoin

public IDataFrame innerJoin(IDataFrame other,
 String leftOn,
 String rightOn)

innerJoin

public IDataFrame innerJoin(IDataFrame other,
 List<String> leftOn,
 List<String> rightOn)

leftJoin

public IDataFrame leftJoin(IDataFrame other,
 String... on)

leftJoin

public IDataFrame leftJoin(IDataFrame other,
 List<String> on)

leftJoin

public IDataFrame leftJoin(IDataFrame other,
 String leftOn,
 String rightOn)

leftJoin

public IDataFrame leftJoin(IDataFrame other,
 List<String> leftOn,
 List<String> rightOn)

rightJoin

public IDataFrame rightJoin(IDataFrame other,
 String... on)

rightJoin

public IDataFrame rightJoin(IDataFrame other,
 List<String> on)

rightJoin

public IDataFrame rightJoin(IDataFrame other,
 String leftOn,
 String rightOn)

rightJoin

public IDataFrame rightJoin(IDataFrame other,
 List<String> leftOn,
 List<String> rightOn)

outerJoin

public IDataFrame outerJoin(IDataFrame other,
 String... on)

outerJoin

public IDataFrame outerJoin(IDataFrame other,
 List<String> on)

outerJoin

public IDataFrame outerJoin(IDataFrame other,
 String leftOn,
 String rightOn)

outerJoin

public IDataFrame outerJoin(IDataFrame other,
 List<String> leftOn,
 List<String> rightOn)

crossJoin

public IDataFrame crossJoin(IDataFrame other)

asOfJoin

public IDataFrame asOfJoin(IDataFrame right,
 String on)

asOfJoin

public IDataFrame asOfJoin(IDataFrame right,
 String on,
 AsOfDirection direction)

asOfJoin

public IDataFrame asOfJoin(IDataFrame right,
 String on,
 String by)

asOfJoin

public IDataFrame asOfJoin(IDataFrame right,
 String on,
 String by,
 AsOfDirection direction)

asOfJoin

public IDataFrame asOfJoin(IDataFrame right,
 String on,
 List<String> by,
 AsOfDirection direction)

asOfJoin

public IDataFrame asOfJoin(IDataFrame right,
 String leftOn,
 String rightOn,
 String leftBy,
 String rightBy,
 AsOfDirection direction)

asOfJoin

public IDataFrame asOfJoin(IDataFrame right,
 String leftOn,
 String rightOn,
 List<String> leftBy,
 List<String> rightBy,
 AsOfDirection direction,
 String leftSuffix,
 String rightSuffix)

concat

public IDataFrame concat(IDataFrame other)

Concatenates this DataFrame with another DataFrame along rows (vertical concatenation) using outer join semantics.

Parameters:

other - DataFrame to concatenate with

Returns:

a new concatenated DataFrame

concat

public IDataFrame concat(IDataFrame other,
 ConcatAxis axis)

Concatenates this DataFrame with another DataFrame along the specified axis using outer join semantics.

Parameters:

other - DataFrame to concatenate with

axis - axis along which to concatenate (ROWS or COLUMNS)

Returns:

a new concatenated DataFrame

concat

public IDataFrame concat(IDataFrame other,
 int axis)

Concatenates this DataFrame with another DataFrame along the specified axis (0 for ROWS, 1 for COLUMNS) using outer join semantics.

Parameters:

other - DataFrame to concatenate with

axis - 0 for ROWS, 1 for COLUMNS

Returns:

a new concatenated DataFrame

concat

public IDataFrame concat(IDataFrame other,
 JoinType how)

Concatenates this DataFrame with another DataFrame along rows using the specified join type.

Parameters:

other - DataFrame to concatenate with

how - join type (OUTER or INNER)

Returns:

a new concatenated DataFrame

concat

public IDataFrame concat(IDataFrame other,
 String how)

Concatenates this DataFrame with another DataFrame along rows using the specified join type name ("outer" or "inner").

Parameters:

other - DataFrame to concatenate with

how - join type name ("outer", "inner")

Returns:

a new concatenated DataFrame

concat

public IDataFrame concat(IDataFrame other,
 ConcatAxis axis,
 JoinType how)

Concatenates this DataFrame with another DataFrame along the specified axis using the given join type.

Parameters:

other - DataFrame to concatenate with

axis - axis along which to concatenate

how - join type (OUTER or INNER)

Returns:

a new concatenated DataFrame

concat

public IDataFrame concat(IDataFrame other,
 int axis,
 JoinType how)

Concatenates this DataFrame with another DataFrame along the specified axis using the given join type.

Parameters:

other - DataFrame to concatenate with

axis - 0 for ROWS, 1 for COLUMNS

how - join type (OUTER or INNER)

Returns:

a new concatenated DataFrame

concat

public IDataFrame concat(IDataFrame other,
 int axis,
 String how)

Concatenates this DataFrame with another DataFrame along the specified axis using the given join type name.

Parameters:

other - DataFrame to concatenate with

axis - 0 for ROWS, 1 for COLUMNS

how - join type name ("outer", "inner")

Returns:

a new concatenated DataFrame

concat

public IDataFrame concat(IDataFrame other,
 ConcatAxis axis,
 String how)

Concatenates this DataFrame with another DataFrame along the specified axis using the given join type name.

Parameters:

other - DataFrame to concatenate with

axis - axis along which to concatenate

how - join type name ("outer", "inner")

Returns:

a new concatenated DataFrame

concat

public IDataFrame concat(IDataFrame other,
 boolean ignoreIndex)

Concatenates this DataFrame with another DataFrame along rows with optional index resetting.

Parameters:

other - DataFrame to concatenate with

ignoreIndex - whether to reset the index

Returns:

a new concatenated DataFrame

concat

public IDataFrame concat(IDataFrame other,
 ConcatAxis axis,
 JoinType how,
 boolean ignoreIndex)

Concatenates this DataFrame with another DataFrame with full control over axis, join type, and index resetting.

Parameters:

other - DataFrame to concatenate with

axis - axis along which to concatenate

how - join type

ignoreIndex - whether to reset the index

Returns:

a new concatenated DataFrame

concatHorizontal

public IDataFrame concatHorizontal(IDataFrame other,
 String leftSuffix,
 String rightSuffix)

Concatenates this DataFrame with another DataFrame horizontally with custom suffixes for colliding columns.

Parameters:

other - DataFrame to concatenate with

leftSuffix - suffix for colliding columns from this DataFrame

rightSuffix - suffix for colliding columns from other DataFrame

Returns:

a new concatenated DataFrame

concatWith

public IDataFrame concatWith(IDataFrame... others)

Concatenates this DataFrame with multiple other DataFrames along rows.

Parameters:

others - other DataFrames to concatenate

Returns:

a new concatenated DataFrame

concatWith

public IDataFrame concatWith(List<? extends IDataFrame> others)

Concatenates this DataFrame with a list of other DataFrames along rows.

Parameters:

others - other DataFrames to concatenate

Returns:

a new concatenated DataFrame

concat

public static IDataFrame concat(List<? extends IDataFrame> frames)

concat

public static IDataFrame concat(List<? extends IDataFrame> frames,
 ConcatAxis axis)

concat

public static IDataFrame concat(List<? extends IDataFrame> frames,
 int axis)

concat

public static IDataFrame concat(List<? extends IDataFrame> frames,
 JoinType how)

concat

public static IDataFrame concat(List<? extends IDataFrame> frames,
 ConcatAxis axis,
 JoinType how)

concat

public static IDataFrame concat(List<? extends IDataFrame> frames,
 ConcatAxis axis,
 JoinType how,
 boolean ignoreIndex)

concat

public static IDataFrame concat(IDataFrame... frames)

concat

public static IDataFrame concat(ConcatAxis axis,
 IDataFrame... frames)

concat

public static IDataFrame concat(int axis,
 IDataFrame... frames)

setHeaders

public void setHeaders(ArrayList<Column> headers)

write

public DataFrameWriter write()

rolling

public RollingWindow rolling(int windowSize)

expanding

public ExpandingWindow expanding()

expanding

public ExpandingWindow expanding(int minPeriods)

window

public Window window()

cumsum

public abstract IDataFrame cumsum(String... columns)

cumprod

public abstract IDataFrame cumprod(String... columns)

cummin

public abstract IDataFrame cummin(String... columns)

cummax

public abstract IDataFrame cummax(String... columns)

rolling_mean

public IDataFrame rolling_mean(String column,
 int windowSize,
 String outputColumn)

rolling_mean

public IDataFrame rolling_mean(String column,
 int windowSize)

rolling_sum

public IDataFrame rolling_sum(String column,
 int windowSize,
 String outputColumn)

rolling_sum

public IDataFrame rolling_sum(String column,
 int windowSize)

lead

public abstract IDataFrame lead(String column,
 int offset,
 String outputColumn)

lead

public IDataFrame lead(String column,
 int offset)

lead

public IDataFrame lead(String column)

lag

public abstract IDataFrame lag(String column,
 int offset,
 String outputColumn)

lag

public IDataFrame lag(String column,
 int offset)

lag

public IDataFrame lag(String column)

diff

public abstract IDataFrame diff(String column,
 int periods,
 String outputColumn)

diff

public IDataFrame diff(String column,
 int periods)

diff

public IDataFrame diff(String column)

diff

public abstract IDataFrame diff(String... columns)

pctChange

public abstract IDataFrame pctChange(String column,
 int periods,
 String outputColumn)

pctChange

public IDataFrame pctChange(String column,
 int periods)

pctChange

public IDataFrame pctChange(String column)

pctChange

public abstract IDataFrame pctChange(String... columns)

rank

public abstract IDataFrame rank(String column,
 String outputColumn)

rank

public IDataFrame rank(String column)

denseRank

public abstract IDataFrame denseRank(String column,
 String outputColumn)

denseRank

public IDataFrame denseRank(String column)

percentRank

public abstract IDataFrame percentRank(String column,
 String outputColumn)

percentRank

public IDataFrame percentRank(String column)

rowNumber

public abstract IDataFrame rowNumber(String outputColumn)

rowNumber

public IDataFrame rowNumber()

plot

public Plot plot()

Obtains a declarative plotting builder for visualizing this DataFrame. Requires the optional cupoftea-plot module on the classpath.

Returns:

a Plot instance

plot

public Plot plot(String engine)

Obtains a declarative plotting builder backed by a specific plotting engine.

Parameters:

engine - plotting engine name (e.g. "vegalite")

Returns:

a Plot instance

toHtml

public String toHtml()

Renders this DataFrame as a clean, responsive HTML table with sticky headers, data type indicators, and client-side pagination.

Returns:

HTML string representation

toHtml

public String toHtml(int maxRows)

Renders this DataFrame as an HTML table displaying up to maxRows rows.

Parameters:

maxRows - maximum number of rows to include

Returns:

HTML string representation

toHtml

public String toHtml(TableHtmlConfiguration configuration)

Renders this DataFrame as an HTML table using the specified configuration.

Parameters:

configuration - table HTML rendering configuration

Returns:

HTML string representation

renderHtml

public String renderHtml()

Jupyter / IJava notebook MIME representation method. Automatically invoked by Jupyter kernels when this DataFrame is evaluated in a cell.

Returns:

HTML table string

render

public Map<String,String> render()

Jupyter / Polyglot notebook MIME representation bundle.

Returns:

map mapping MIME type ("text/html") to HTML representation