search

ExposuresProvider / icees-api

MIT License
2 stars 8 forks source link

Bonferroni correction for "association to all features" #85

Open karafecho opened 3 years ago

karafecho commented 3 years ago

This issue is to report presumed inaccuracies with the Bonferroni correction that is applied to the ICEES+ "association to all features" functionality. Specifically, (1) the correction appears to be a mandatory parameters (it was optional); (2) some sort of correction is applied even when the maximum_p_value is set to 1; and (3) users should be required to set the alpha value, instead of or in addition to the maximum_p_value. The result is that many associations are being filtered out, resulting in what appears at first glance to be missing variables.

It is possible that this issue explains some of the other recent issues that have been posted.

karafecho commented 3 years ago

So, I found a workaround solution to this issue, but it doesn't make sense. If I set the maximum_p_value to 10, I receive a full set of associations (I think). However, a P value of 10 has no logical interpretation.