Window Transform
Note
Transforms are a part of the underlying language, which is not directly accessible to users. This content is maintained for reference purposes only. For more information on the user-accessible equivalent to transforms, see Transformation Reference.
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.
For more information on window functions, see Window Functions.
You can also use the aggregation functions with this transform. See Aggregate Functions.
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.
Basic Usage
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.
Syntax and Parameters
window value: WINDOW_FUNCTION(arg1,arg2) order: order_col [group: group_col]
Token | Required? | Data Type | Description |
|---|---|---|---|
window | Y | transform | Name of the transform |
value | Y | string | Expression that evaluates to the window function call and its parameters |
order | Y | string | Column or column names by which to sort the dataset before the |
group | N | string | Column name or names containing the values by which to group for calculation |
For more information on syntax standards, see Language Documentation Syntax 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.
Usage Notes:
Required? | Data Type |
|---|---|
Yes | String (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.
Note
The order parameter must unambiguously specify an ordering for the data, or the generated results may vary between job executions.
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.
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.
Note
The order column does not need to be sorted before the window transform is executed on it.
Astuce
To sort in reverse order, prepend the column name with a dash (-MyDate).
Usage Notes:
Required? | Data Type |
|---|---|
Yes | String (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.
Note
Transforms that use the group parameter can result in non-deterministic re-ordering in the data grid. However, you should apply the group parameter, particularly on larger datasets, or your job may run out of memory and fail. To enforce row ordering, you can use the sort transform. For more information, see Sort Transform.
Usage Notes:
Required? | Data Type |
|---|---|
No | String (column name) |
Examples
Astuce
For additional examples, see Common Tasks.
See the individual functions for examples. See Window Functions.