Feature engineering with Python

category coding (FE cookbook, 2nd ed (Packt))

One-hot encoding
OHE - frequent categories
Categories -> frequency counts
ordinals
ordinals (ordered)
target mean encoding
weight of evidence encoding
grouping rare categories
binary encoding

composite estimators (transformers) (scikit-learn 0.24)

setup, tips, caching, regression target transforms

creation (FE cookbook, 2nd ed (Packt))

combining features
comparing features to reference values
polynomial expansion
combining features with trees
periodic data
splines
features|figures (splines)
features|figures (cyclical)
features|figures (polynomial)

data imputation (FE cookbook, 2nd ed (Packt))

store credit approval dataset
missing data -> removal
missing data -> means or medians
categorical data
missing data -> arbitrary values
finding extreme values
marking imputed values
multivariate (chained) imputation
estimated imputation - K nearest neighbors

data imputation basics (scikit-learn 0.24)

univariate, multivariate, nearest-neighbor, marking imputed values

data transforms (FE cookbook, 2nd ed (Packt))

logarithmic
reciprocal
square root
power
box-cox
yeo-johnson

datasets - simple examples (scikit-learn 0.24)

iris, digits, cal housing, labeled faces, 20 newsgroups, (more)

date/time handling (FE cookbook, 2nd ed (Packt))

date feature extraction
time feature extraction
capturing elapsed time
time zones
automation with feature engine

discretization (binning) (FE cookbook, 2nd ed (Packt))

equal-width
equal-frequency
user-defined widths
discretization with k-means
binarization
using decision trees

feature engineering intro (python DS handbook)

one-hot encoding, word counts, tf-idf, linear-to-polynomial, missing data, pipelines

feature extraction (text) (scikit-learn 0.24)

bag of words, sparsity, vectorizers, stop words, tf-idf, decoding, applications, limits, the hashing trick, out-of-core ops

file i/o (numeric python)

CSV, HDF5, h5py, pytables, hdfstore, JSON, serialization, pickle issues

outlier management (FE cookbook, 2nd ed (Packt))

visualization with boxplots
finding with mean & stdev
finding with IQR
removal
capping
capping with quantiles

preprocessing basics (scikit-learn 0.24)

mean removal, variance scaling, sparse scaling, outlier scaling, distribution maps, normalization, category coding, binning, binarization, polynomial features.

random projections (scikit-learn 0.24)

scaling (FE cookbook, 2nd ed (Packt))

standardization
min-max scaling
robust scaling
mean normalization
max absolute scaling
unit-length scaling

tools (featuretools) (FE cookbook, 2nd ed (Packt))

entity set setup
creating features - cumulative primitives
combining numeric features
creating features - datetimes
creating features - aggregations
retail dataset

tools (tsfresh) (FE cookbook, 2nd ed (Packt))

auto extraction
relevant feature extraction
specific feature extraction
post-selection feature extraction
occupancy dataset

Perfectly Awesome

Feature Engineering with Python