From c24ce7c891934853e89af171d8104503a6f958f0 Mon Sep 17 00:00:00 2001 From: "Maja Karwowska (Kabus)" <8majak8@gmail.com> Date: Tue, 18 Mar 2025 17:51:44 +0100 Subject: [PATCH] Fix wrong example for sliding windows --- docs/basics-tasks/CombiningData.md | 3 +-- 1 file changed, 1 insertion(+), 2 deletions(-) diff --git a/docs/basics-tasks/CombiningData.md b/docs/basics-tasks/CombiningData.md index 6c2a38ae..cef000b2 100644 --- a/docs/basics-tasks/CombiningData.md +++ b/docs/basics-tasks/CombiningData.md @@ -102,8 +102,7 @@ Block policies have 2 additional parameters: `outsider` and `categoryNeighbours` `CategoryNeighbours` is the number of the consecutive following elements a given element is combined with. For performance reasons, tuples of elements traditionally are not generated over the whole bin, but over several much smaller intervals.
Example: `categoryNeighbours = 4`, the bin contains elements at rows: 1, 3, 5, 6, 10, 13, 16, 19
- Strictly upper pairs (different colors mark pairs from different 5-element combinations intervals): (1, 3), ..., (1, 10), (3, 5), ..., (6, 10), (3, 5), ..., (10, 13), (5, 6), ..., (13, 16), ...
-Note that some pairs get repeated, e.g., (3, 5).
+ Strictly upper pairs (different colors mark pairs from different 5-element combinations intervals): (1, 3), ..., (1, 10), (3, 5), ..., (3, 13), (5, 6), ..., (5, 16), ...
To get the behavior without sliding windows, set category neighbours to a very high value. Below, you can see a full example of block combinations in an analysis task: