Pre-processing steps
The following pre-processing methods are included in autoMLk, with their respective hyper-parameters:
categorical encoding:
- No encoding
- **
- Label Encoder
- **
- One hot categorical
- drop_invariant
- BaseN categorical
- drop_invariant, base
- Hashing categorical
- drop_invariant
text encoding:
Bag of words
Word2Vec
Doc2Vec
imputing missing values:
- No missing
- **
- Missing values fixed
- fixed
- Missing values frequencies
- frequency
feature scaling:
- No scaling
- **
- Scaling Standard
- **
- Scaling MinMax
- **
- Scaling MaxAbs
- **
- Scaling Robust
- quantile_range
feature selection:
- No Feature selection
- **
- Truncated SVD
- n_components, algorithm
- Fast ICA
- n_components, algorithm
- PCA
- n_components
- Selection RF
- n_estimators
- Selection RF
- n_estimators
- Selection LSVR
- **