Increase Kotlin duplication mass threshold

[larkinscott]

• After looking at some of the duplication issues this was surfacing locally and running some metrics, I think it would be good to bump this from 40 to 60

[chrishulton]

Seems fine, just interested how you came up with 60 (and not say, 50, 70, etc).

[larkinscott]

@chrishulton I was leaning towards 60 after looking at a range of the issues that were being surfaced, so I decided to try to get some metrics. It's not perfect, but I took a handful of Kotlin repos, dropped them into a directory, set the mass_threshold to 20 and collected all the resulting mass_thresholds using this: https://github.com/codeclimate/qm_qa/pull/16/files

The result was:

{'metrics': [{'language': 'Kotlin', '90th percentile': 84.0, '95th percentile': 126.0, 'Mean': 55.250245076196414, 'Average': 55.250245076196414, 'Median': 39.0, 'Standard deviatation': 80.4210381927305, 'Max value': 2777, 'Min value': 20}]}

I think the mean seems close here, so I ended up going with my first impression which was 60. Anything higher would eliminate some actual duplication issues.

[chrishulton]

Cool, I'm not really sure what kind of math makes sense here but that at least sounds reasonable.

If we had good issue status data (mark as confirmed / mark as invalid) this could be a good candidate for machine learning. Maybe something we can try out later ..........