Closed YoshikawaMasashi closed 5 years ago
Merging #318 into master will decrease coverage by
5.67%
. The diff coverage is90.62%
.
@@ Coverage Diff @@
## master #318 +/- ##
==========================================
- Coverage 89.59% 83.91% -5.68%
==========================================
Files 182 184 +2
Lines 8272 8432 +160
==========================================
- Hits 7411 7076 -335
- Misses 861 1356 +495
Is the implementation follow from some paper? If so, please cite inside docstring
.
I also wonder BaseScaler
should have inverse_transform
method or not. (which means, this scaler should be BaseScaler or other class maybe Normalizer class?)
The scaler without inverse_transform
is still useful for input feature normalization, but not applicable for label scaling.
LGTM except that.
I am implementing "Flow Scaler", that scale data into the normal distribution. This Scaler uses a technique named "flow".