DBSCAN: Optional normalization

[VesnaT]

Issue

Fixes #5424

Description of changes

Add a check box to normalize/not normalize the input dataset. The check box is checked by default.

Includes

• Code changes
• Tests
• Documentation

[codecov]

Codecov Report

Merging #5428 (5a0f03e) into master (ce6222e) will decrease coverage by 0.00%.
The diff coverage is 100.00%.

@@            Coverage Diff             @@
##           master    #5428      +/-   ##
==========================================
- Coverage   86.14%   86.14%   -0.01%     
==========================================
  Files         303      303              
  Lines       61592    61607      +15     
==========================================
+ Hits        53061    53072      +11     
- Misses       8531     8535       +4     

[markotoplak]

@VesnaT, thanks, looks good to me!

[borondics]

This is a nice addition, but what about the other preprocessors in DBSCAN? The user still doesn't know about them, even if they are necessary...

The GUI across widgets is not consistent. I like k-means as there is a Preprocessing group, so one could add everything there and in DBSCAN's case the Distance metric could continue having its own group.

[markotoplak]

The other preprocessors are necessary if you do not want the widget to work at all with missing values or some discrete columns. If your data does not have either they will not do anything anyway.

The UI's are different because one allows distance measurement method setting and the other does not.

[janezd]

But preprocessors should be listed in documentation, I guess. For this widget and for others.

[irgolic]

Could we automate/standardize preprocessors, to make them a bit more transparent?