Philipp Probst
Philipp Probst
Did you consider using more datasets? And how about regression problems? There is for example this benchmarking suite, accessible via the OpenML packages: [https://arxiv.org/abs/1708.03731](https://arxiv.org/abs/1708.03731)
I made a benchmark comparing it with `ranger` and `tuneRanger` in the default mode on some regression datasets (surprisingly `tuneRanger` was quite good, maybe datasets too small?) and noticed, that...
Did you consider in making (also) an autocatboost algorithm? On their webpage they claim to be better regarding performance and runtime: [https://catboost.yandex/](https://catboost.yandex/) Would be interesting to compare this properly. Unfortunately...
Dear @boxuancui, this is a nice package. It would be nice if there would be a default option for producing plotly plots instead of simple ggplot2 plots. Currently you can...
I got worse results for liquidSVM in a regression benchmark compared to the benchmark that I did around one year ago. Can there be specific reason for this, did you...
I think it would be cool to have a package like this, but just for the explorative data analysis. Click on some variables, look at the associations and maybe save...
We could order the learners due to their performance and give somehow some recommendations on them.
Would be cool to have a time estimation in advance after choosing a learner and a progress bar, so you see how long it still takes to train/benchmark your model.
I tried out Lick with Puppy Linux and it was great. Now I cleaned up again, but I cannot enable "fast Windows start" again. What do I have to do...
I think it could be interesting to introduce the concept of "effect size" in chapter 5.9: https://novustat.com/statistik-blog/effektstaerke-berechnen-beta-koeffizient.html Because the raw value is a very unstandardized way of evaluating the effect...