You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
* Added Ranger, XGBoost, fastNaiveBayes and LiblineaR high performing algorithms
6
+
* Added the autoRLearn_ function, which assumes that the data is in perfect shape and can be loaded from a dataframe, unlike autoRLearn which can only load from a data file outside R.
7
+
* Added Hyperband and Bayesian Optimization Hyperband to the new autoRLearn_
8
+
* Added some extra temporary dependencies which will be removed in the following months (all tidyverse packages other than purrr)
9
+
* Fixed some small mistakes in the code and jsons
10
+
11
+
## Current Roadmap
12
+
13
+
* fix metalearning, at the moment it doesn't work. There's something wrong with the AWS server we are using.
14
+
* change the dplyr back end to use data.table with dtplyr
15
+
* merge autoRLearn and autoRLearn_ into a single function, which can both load from a data file and in R.
16
+
* Rewrite SMAC, as requested by Sherif.
17
+
18
+
## Extra info
19
+
20
+
* brurucy is a new and active maintainer
21
+
* Nightly and experimental versions, independent from the Data Systems Lab, are being developed at https://github.com/brurucy/witchcraft
22
+
* Updates will be conservative and focused on non-breaking changes, up to release 1.0.
0 commit comments