Update README.md

README.md

@@ -1212,7 +1212,7 @@ The dataset presents eleven sections (one per regression task), with varying deg
 
 The sections are already partitioned into a predefined train-validation-test split for training assessors. Assessors need a particular kind of partitioning (mainly stratified by instance identifier to avoid contamination), so that's why the subsets are given. 
 
-The **main** branch contains the unaltered datasets, keeping the original values of the task and model characteristics, whereas the **normalised** branch contains the datasets properly normalised.
+The **main** branch contains the unaltered datasets, keeping the original values of the task and model characteristics, whereas the **normalised** branch contains the datasets properly normalised (numerical features are centered and scaled and categorical features are transformed into dummies).
 
 ## Original tasks