What are the Qlik Sense Scalability Tool's Regression Analyzer's limitations in terms of finding differences between two regression logs?
I have two regression logs where the main difference is data displayed in pivot tables. One table's values is 1.1 times the other.
When I run the Regression Analyzer to compare the logs, it returns zero differences. When I go into Advanced view and choose the pivot table from the object list on the left, it seems like the tool does have some inkling that there's a difference because the numbers are highlighted in grey (screenshot attached). But is there a way for the tool to highlight that difference straight away?
I wouldn't want to have to scroll thru all the captured results just to make sure no numbers are greyed out...
Hello! That is unfortunate. We were not previously aware of issues with identifying the data in pivot tables but this might be a bug. I will look further into diffing pivot tables with the regression analyzer.
It's hard to tell without data, and I appreciate that your data is confidential, but it would be most helpful if you would have anything to share which could help reproduce the issue.
Looking at the source code and trying to figure out what's going on, one suspicion I have is that we might be missing diffs that are within a qPivotDataPages. Is that the case in your analysis?
Hi! I did not set up the app myself, but from what I can tell reading the regression results, yes they do appear to be within qPivotDataPages.
And yes, I'm not really at liberty to share all the logs as there is some confidential data involved.
One workaround I found for now is that ticking the "Condensed view" box lets me see these greyed out values straight away (screenshot attached). So perhaps it's more a matter of having the tool detect these variances as significant differences (rather than negligible ones) more than anything else...
Thank you for confirming. Yes I think a change to the rules that determine what is and isn't an important diff will produce the expected results. I believe I have identified the issue but I will need some time to reproduce and verify it. I will get get back to you then. Again, thanks for bringing this issue to our attention.