ResampleWithMissingProcessor.h

#ifndef TRAIN_WITH_MISSING_PROCESSOR_H__
#define TRAIN_WITH_MISSING_PROCESSOR_H__
 
#include "FeatureProcess.h"
 
class ResampleMissingProcessor : public FeatureProcessor {
public:
    vector<string> selected_tags; 
    vector<string> removed_tags; 
    float missing_value; 
    bool duplicate_only_with_missing; 
 
    //grouping of imputattion - for example handle imputations by signal (or other groups):
    string grouping; 
 
    int add_new_data; 
    //Sampling options - general parameters:
    int limit_mask_size; 
    bool sample_masks_with_repeats; 
    //Sampling random
    bool uniform_rand; 
    float uniform_rand_p; 
    //When Sampling not random uniform_rand is false
    bool use_shuffle; 
 
    int subsample_train; 
    bool verbose; 
 
    string select_learn_matrix(const vector<string> &matrix_tags) const;
 
    ResampleMissingProcessor() : FeatureProcessor() { init_defaults(); }
    // Copy
    virtual void copy(FeatureProcessor *processor) { *this = *(dynamic_cast<ResampleMissingProcessor *>(processor)); }
 
    int init(map<string, string>& mapper);
    void init_defaults();
 
    //print function
    void dprint(const string &pref, int fp_flag);
 
    // Apply do nothing in apply
    int _apply(MedFeatures& features, unordered_set<int>& ids) { return 0; }
    int Learn(MedFeatures& features, unordered_set<int>& ids);
 
    // Serialization
    ADD_CLASS_NAME(ResampleMissingProcessor)
        ADD_SERIALIZATION_FUNCS(processor_type, selected_tags, removed_tags, missing_value, add_new_data, sample_masks_with_repeats,
            uniform_rand, uniform_rand_p, use_shuffle, subsample_train, limit_mask_size, grouping, groupNames,
            group2Inds, verbose, duplicate_only_with_missing)
private:
    vector<vector<int>> group2Inds;
    vector<string> groupNames;
};
 
MEDSERIALIZE_SUPPORT(ResampleMissingProcessor)
 
#endif // !TRAIN_WITH_MISSING_PROCESSOR_H__