Transformation Operators

Transformation operators provide different ways to prepare data for modeling.

  • Aggregation (DB)
    Performs aggregate calculations on a data set by specifying a group-by configuration and an aggregate expression.
  • Aggregation (HD)
    Performs aggregate calculations on a data set by specifying a group-by configuration and an aggregate expression.
  • Batch Aggregation
    Performs aggregations on multiple columns.
  • Collapse
    Transforms the data contained in a column of a table by means of subtotals (or other calculations) that are defined by another column in the same list. The other calculations might be averages and counts. The result is a collapsed or condensed data set.
  • Column Filter (DB)
    Selects a subset of the columns from data source. Only the columns selected remain in the output dataset.
  • Column Filter (HD)
    Selects a subset of the columns from data source. Only the columns selected remain in the output dataset.
  • Correlation Filter (DB)
    Filters numeric columns so the remaining columns are not correlated strongly with each other.
  • Correlation Filter (HD)
    Filters numeric columns so the remaining columns are not correlated strongly with each other.
  • Distinct (DB)
    Returns only distinct combinations of values from specified columns of a database source. Rows are not returned in any particular order, but each combination of values within a row is distinct from other rows.
  • Distinct (HD)
    Returns the unique value combinations across selected columns.
  • Fuzzy Join
    Performs a fuzzy matching join to connect two data sets based on nearly matching string values.
  • Join (DB)
    Performs a table join on the input data sets by allowing users to define the input data set alias, the output columns, and the join condition.
  • Join (HD)
    Performs a table join on the input data sets by allowing users to define the input data set alias, the output columns, and the join condition.
  • Normalization (DB)
    Performs normalization on the selected columns of the input data set. Normalization means adjusting values measured on different scales to a notionally common scale.
  • Normalization (HD)
    Performs normalization on the selected columns of the input data set. Normalization means adjusting values measured on different scales to a notionally common scale.
  • Null Value Replacement (DB)
    Replaces null values of the selected fields of the data set with designated values. This is helpful as a pre-cleansing data step.
  • Null Value Replacement (HD)
    Replaces null values of the selected fields of the data set with designated values. This is helpful as a pre-cleansing data step.
  • Numeric to Text (DB)
    Converts a numeric type column to a text type column.
  • Numeric to Text (HD)
    Converts a numeric type column to a text type column.
  • One-Hot Encoding
    Performs one-hot encoding on a set of categorical columns selected: it encodes categorical features using a one-hot scheme (also known as "one-of-K" scheme), and outputs a binary column for each distinct category in the input column.
  • Pivot (DB)
    Lets you transform the categorical data contained in a column of a table into columns of a new table, by means of subtotals (or other calculations) that might be defined by another column in the same list. The other calculations might be averages and counts.
  • Pivot (HD)
    Lets you transform the categorical data contained in a column of a table into columns of a new table, by means of subtotals (or other calculations) that might be defined by another column in the same list. The other calculations might be averages and counts.
  • Reorder Columns (DB)
    Reorders one or more columns from an input table, and optionally renames them.
  • Reorder Columns (HD)
    Reorders one or more columns from an input table, and optionally renames them.
  • Replace Outliers (DB)
    Reduces the range of values for numeric columns.
  • Replace Outliers (HD)
    Reduces the range of values for numeric columns.
  • Row Filter (DB)
    Sets the criteria for filtering data set rows. Only the rows that meet the criteria remain in the output data set.
  • Row Filter (HD)
    Sets the criteria for filtering data set rows. Only the rows that meet the criteria remain in the output data set.
  • Sessionization
    Enables the application of sessionization on time-series data to create a session_id column that, for each row (and user ID), gives the session the action belongs to.
  • Set Operations (DB)
    Combines results from merging two or more queries into a single result set.
  • Set Operations (HD)
    Combines results from merging two or more queries into a single result set.
  • Sort By Multiple Columns
    Allows you to choose up to three columns to sort by and returns a data set sorted by the selected column(s), adding a column called row_index that enables you to filter the output based on the sorting results.
  • Transpose
    Allows you to rearrange data so that rows and columns are switched.
  • Unpivot (DB)
    Unpivots one or more columns.
  • Unpivot (HD)
    Unpivots one or more columns.
  • Unstack
    Takes an HDFS data set in stacked format and produces an unstacked (wide) HDFS data set using user-specified grouping and pivot columns.
  • Variable (DB)
    Use to define variables created from data fields of the input data set, forming a new table or view.
  • Variable (HD)
    Use to define variables created from data fields of the input data set, forming a new table or view.
  • Window Functions - Aggregate
    Unlike regular aggregate functions calls, allows you to create aggregate variables for each input row, based on the specified frame (with an optional order).
  • Window Functions - Lag/Lead
    For several columns and offset values (n), returns the value of the column that is n rows before (lag) or after (lead) the current row.
  • Window Functions - Rank
    Returns the rank of each row in relation to its windowed partition.
Related concepts