I have a dataset with all the features and the required output with categorical values. I would like to select combinations(multiple - can be as high as over 100 combinations) of features (in order), that best describes the final output. Are there any python library I can use for this situation? Else, please suggest a method. The final sets of features(+values) --> output can may be considered as rules to describe the output.\
| X1 | X2 | X3 | Y |
|---|---|---|---|
| a1 | b2 | c1 | y1 |
| a2 | b3 | c2 | y2 |
| a1 | b2 | c2 | y1 |
| a2 | b3 | c3 | y2 |
Some possible combinations of features - {C1=a1,C2=b2,C3=c1 --> y=y1} ; {C2=b2, C1=a1, C3=C1 --> y=y1} ; {C1=a1, C2=b2 --> y=y1} ; {C2=b3,C1=a2 --> y2}
Tried decision trees and path-constrained random walk as some methods for a starting point, but they aren't feasible for large sets of feature-value(s) combinations. \ Need a faster and simpler generalization method to generate the final combinations.
