[Bug]: PC scores added on wrong row if data have pass-through filters #2972
Description
JASP Version
0.19.1
Commit ID
No response
JASP Module
Factor
What analysis are you seeing the problem on?
PCA
What OS are you seeing the problem on?
Windows 10
Bug Description
When performing a PCA and selecting "add PC scores to data", the scores are inserted correctly if no observations are filtered out. However, if any observations in the data are filtered out, then data in the PC score columns will be added incorrectly. Specifically, if you filter out F observations, the software is clever enough to know it needs to generate N-F scores, but then when adding the PC score column in the data it doesn't skip the filtered out rows (which means the score are translated by F rows).
This is easy to see in the following figures:
FIGURE 1: 29 Observations, Filtered: 0; all PC scores added correctly
FIGURE 2: 29 Observations, Filtered: 1 (1st row); scores added incorrectly (i.e., PC score was added to the filtered observation, which should have none, and missing in the last row). In short, all scores are added to the wrong row, by 1 position.
FIGURE 3: 29 Observations, Filtered: 2 (1st and 2nd rows); scores added incorrectly (i.e., PC score was added to the filtered observation, which should have none, and missing in the last row). In short, all scores are added to the wrong row, by 2 positions.
Of course, the same is true if you filter out rows from the middle of the data. From the first filtered out observation down, all the scores are offset.
So of course, if then you use those scores to do any additional analysis, the result will be incorrect (since each score is in the wrong row).
I hope I was able to explain clearly enough, should be easy to reproduce. Let me know what you think
Cheers
Martin
Expected Behaviour
When inserting the PC score column(s) it should skip any filtered-out row.
Steps to Reproduce
- Perform a PCA on full dataset, selecting "add PC scores to data"
- Apply a pass-through filter which filters out one or more observations
- Look at the data again, you should see as many empty cells, per each PC, as filtered out observations (refreshing or re-running the PCA does not change the behavior)
...
Log (if any)
No response
More Debug Information
No response
Final Checklist
- I have included a screenshot showcasing the issue, if possible.
- I have included a JASP file (zipped) or data file that causes the crash/bug, if applicable.
- I have accurately described the bug, and steps to reproduce it.