Perl 6 - the future is here, just unevenly distributed

IRC log for #bioclipse, 2014-06-04

| Channels | #bioclipse index | Today | | Search | Google Search | Plain-Text | summary

All times shown according to UTC.

Time Nick Message
05:11 egonw joined #bioclipse
06:05 zaetnick joined #bioclipse
06:31 egonw_ joined #bioclipse
07:21 zaetnick joined #bioclipse
07:42 egonw_ joined #bioclipse
07:53 Gpox joined #bioclipse
11:00 zaetnick_ joined #bioclipse
11:11 Pyryhyry joined #bioclipse
11:12 Pyryhyry hi guys! may i ask a chemometrics-related non-bioclipse question? (i attended your Pharmaceutical Bioinformatics web course a few years ago)
11:14 Pyryhyry Random forests have this feature that you can do random permutations to a predictor and then see how big changes that brings to the overall predictors, and this way you can check the relative importance of the predictors, i.e. changing the most important predictors cause the biggest worsening of the predictions
11:15 samuell joined #bioclipse
11:15 Pyryhyry i think this can be calculated by importance(randomForest(...))
11:17 Pyryhyry now, this does not take into account the correlations between the predictors, so the question finally: Do you know of a nice or quick-and-dirty way of selecting the most important predictors to be used e.g. in multiple linear regression (in other words, the predictors also should be uncorrelated)
11:28 zaetnick joined #bioclipse
14:50 zaetnick joined #bioclipse
20:39 jonalv joined #bioclipse
22:03 egonw_ joined #bioclipse

| Channels | #bioclipse index | Today | | Search | Google Search | Plain-Text | summary