Skip to contents

This is the reverse to groupBy(...).pivot(...).agg(...), except for the aggregation, which cannot be reversed.

Usage

unpivot(x, ids, values, variableColumnName, valueColumnName)

# S4 method for SparkDataFrame,ANY,ANY,character,character
unpivot(x, ids, values, variableColumnName, valueColumnName)

# S4 method for SparkDataFrame,ANY,ANY,character,character
melt(x, ids, values, variableColumnName, valueColumnName)

Arguments

x

a SparkDataFrame.

ids

a character vector or a list of columns

values

a character vector, a list of columns or NULL. If not NULL must not be empty. If NULL, uses all columns that are not set as ids.

variableColumnName

character Name of the variable column.

valueColumnName

character Name of the value column.

Value

a SparkDataFrame.

Note

unpivot since 3.4.0

melt since 3.4.0

Examples

if (FALSE) {
df <- createDataFrame(data.frame(
  id = 1:3, x = c(1, 3, 5), y = c(2, 4, 6), z = c(-1, 0, 1)
))

head(unpivot(df, "id", c("x", "y"), "var", "val"))

head(unpivot(df, "id", NULL, "var", "val"))
}