A mix of tools to test and compare:
*Ensembles methods
*Filtering algorithm based on "Cognitive Diversity" to pass to Ensembles methods
*Parameters fitting strategies (Gridsearch,...)
Board_opinion is class that will handle the training of a set of pipelines. It handles variabilities filtering, normalization, dimension reduction, and finally the classifiers themselves. It will generate every combinations of variabilities, normalization , dimension reduction and classifiers given. NOT every combination of parameters just the algos.
Ensemble method:
*New_mgs contains the class MGS for Mixed Group Scores handling a modified soft + beta implementation of the Mixed Group Ranks algo.
*For now other ensembles method are not defined here.
Filtering methods:
Performance: Select x, given number, of clfs based on performance. CDS: Cognitive Diversity Strength as defined in "Preference Prediction Based on Eye Movement Using Multi-layer Combinatorial Fusion". Sliding Ruler: Ranks by performance a separately by CDS before selecting the X highest common to both.