Search

1. Preface
2. Transformations
3. Source transformation
4. Target transformation
5. Aggregator transformation
6. Cleanse transformation
8. Data Services transformation
9. Deduplicate transformation
10. Expression transformation
11. Filter transformation
12. Hierarchy Builder transformation
13. Hierarchy Parser transformation
14. Hierarchy Processor transformation
15. Input transformation
16. Java transformation
17. Java transformation API reference
18. Joiner transformation
19. Labeler transformation
20. Lookup transformation
21. Machine Learning transformation
22. Mapplet transformation
23. Normalizer transformation
24. Output transformation
25. Parse transformation
26. Python transformation
27. Rank transformation
28. Router transformation
29. Rule Specification transformation
30. Sequence Generator transformation
31. Sorter transformation
32. SQL transformation
33. Structure Parser transformation
34. Transaction Control transformation
35. Union transformation
36. Velocity transformation
37. Verifier transformation
38. Web Services transformation

# Frame

The frame determines which rows are included in the calculation for the current input row based on the rows' relative position to the current row. Configure a frame if you use an aggregate function as a window function.
The start offset and end offset describe the number of rows that appear before and after the current input row. An offset of "0" represents the current input row. For example, a start offset of -3 and an end offset of 0 describe a frame including the current input row and the three rows before the current row.
The following image shows a frame with a start offset of -1 and an end offset of 1:
For every input row, the function performs an aggregate operation on the rows inside the frame. If you configure an aggregate expression like SUM on the frame in the previous image, the expression calculates the sum of the values within the frame and returns a value of 6000 for the input row.
You can also specify a frame that does not include the current input row. For example, a start offset of 10 and an end offset of 15 describe a frame that includes six total rows, from the tenth to the fifteenth row after the current row.
Offsets of
All Preceding Rows
and
All Following Rows
represent the first row of the partition and the last row of the partition. For example, if the start offset is All Preceding Rows and the end offset is -1, the frame includes one row before the current row and all rows before that.
The following image shows a frame with a start offset of 0 and an end offset of All Following Rows:
If the frame offsets are outside the partition, the aggregate function ignores the frame. If the offsets of a frame are not within the partition or table, the aggregate function processes only the rows within the partition. The aggregate function lists the skipped rows in the log file. The function returns one of the following responses for the skipped rows: a default value of ERROR('transformation error'), NULL, or a predefined constant.
For example, you partition a table by seller ID and you order by quantity. You set the start offset to -3 and the end offset to 4.
The following image shows the partition and frame for the current input row:
The frame includes eight total rows, but the calculation remains within the partition. If you define an AVG function with this frame, the function calculates the average of the quantities inside the partition and returns 18.75.
Consider the following rules and guidelines when you define a frame:
• LEAD and LAG use the frame that you specify in the function arguments and ignore the frame that you configure on the
Window
tab.
• The start offset must be less than or equal to the end offset.

Updated March 06, 2023

Resources