How to Select Categorical Input Features: Encoding and K-best

Question 1

DOES PANDAS TRY TO MAP SOME STR INPUTS TO NUMERICAL VALUES IN THE DATASET? IF YES, WHAT SHOULD WE DO ABOUT IT?P137

Answer 1

Yes, that’s why it’s better to convert for example, numbers with str format, into str, after reading the dataset file.

Question 2

DOES THE ORDINALENCODER IN SCIKIT-LEARN ALLOW SPECIFYING THE ORDER OF CATEGORIES? P139

Answer 2

Yes, it does.

Question 3

WHAT IS THE DIFFERENCE BETWEEN ORDINALENCODER AND LABELENCODER? P139

Answer 3

Labelencoder is for encoding a single variable, can’t give a dataframe to it

Question 4

WHEN USING K-BEST FOR CHI2 TEST, WHAT DO HIGHER SCORES INDICATE? P141

Answer 4

Stronger dependence between feature and target (it doesn’t show p-value, it’s just a score)

Question 5

HOW CAN WE USE MUTUAL INFO IN K-BEST FOR CLASSIFICATION PROBLEMS? P152

Answer 5

By setting “score_func” parameter to mutual_info_classif