Page tree

 

Support | BlogContact Us | 844.332.2821

Versions Compared

Key

  • This line was added.
  • This line was removed.
  • Formatting was changed.
Comment: Migration of unmigrated content due to installation of a new plugin

D toc
Excerpt

The window transform enables you to perform summations and calculations based on a rolling window of data relative to the current row.  

For example, you can compute the rolling average for a specified column for the current row value and the three preceding rows. This transform is particularly useful for processing time or otherwise sequential data.

You can apply one or more functions to your window transform step. 

Info

NOTE: Be careful applying this transform across a large number of rows. In some cases, the application can run out of memory generating the results, and your results can fail.

D s
snippetBasic

D code
window value: ROLLINGAVERAGE(myValues,3) order: MyDate group: customerId

Output: Generates a new column called, window, which contains the result of the ROLLINGAVERAGE function applied from the current row in the myValues column across the 3 rows forward, ordered by MyDate and grouped by customerId

D s
snippetParameters
D code

window value: WINDOW_FUNCTION(arg1,arg2) order: order_col [group: group_col]

ParameterRequired?Transform BuilderData TypeDescription
valueYFormulastringExpression that evaluates to the window function call and its parameters
orderYOrder bystringColumn or column names by which to sort the dataset before the value expression is applied
groupNGroup bystringColumn name or names containing the values by which to group for calculation

d-s-lang-notes

value

For the window transform, the value parameter contains the function call or calls, which define the set of rows to which the function is applied.

You can specify multiple window functions for the value. Each function reference must be separated by a comma. The transform generates a new column for each window function.

This transform uses a special set of functions. For more information on the available functions, see  Window Functions.

D s
snippetusage

Required?Data Type
YesString (expression)

order

For the window transform, this parameter specifies the column on which to sort the dataset before applying the specified function. For combination sort keys, you can add multiple comma-separated columns.

Info

NOTE: The order parameter only applies to transform steps containing a window function or to the sort transform. For steps that contain a transform supporting an order parameter with a non-window function, the order parameter does nothing.

Info

NOTE: If you are applying a window function, it requires a primary key to identify rows in the output. Otherwise, results can be ambiguous. For more information on defining a primary key, see Window Functions.

Info

NOTE: If it is present, the dataset is first grouped by the group value before it is ordered by the values in the order column.

Info

NOTE: The order column does not need to be sorted before the window transform is executed on it.

Tip

Tip: To sort in reverse order, prepend the column name with a dash (-MyDate).

D s
snippetusage

Required?Data Type
YesString (column name)

group

For the window transform, this parameter specifies the column whose values are used to group the dataset prior to applying the specified function. For combination grouping, you can specify multiple comma-separated column names.

Info

NOTE: Be careful applying this transform across groups containing a large number of rows. In some cases, the application can run out of memory generating the results, and your results can fail.

D s
snippetusage

Required?Data Type
NoString (column name)


D s
snippetExamples

See the individual functions for examples. See Window Functions.

D s also
labelwrangle_transform_window