PredictorImputer.h

#ifndef _FTR_PREDICTOR_IMPUTER_H_
#define _FTR_PREDICTOR_IMPUTER_H_
 
#include <string>
#include <random>
#include <MedAlgo/MedAlgo/BinSplitOptimizer.h>
#include <MedProcessTools/MedProcessTools/Calibration.h>
#include <MedAlgo/MedAlgo/SamplesGenerator.h>
 
using namespace std;
 
class PredictorImputer : public FeatureProcessor {
private:
    unique_ptr<SamplesGenerator<float>> _sampler = NULL;
    void *sampler_sampling_args = NULL;
 
    GibbsSampler<float> _gibbs;
    GibbsSamplingParams _gibbs_sample_params;
    int n_masks = 1;
    vector<string> impute_features;
 
    void init_sampler(bool with_sampler = true);
 
    mt19937 gen;
public:
    float missing_value; 
    bool verbose_learn; 
    bool verbose_apply; 
    string tag_search; 
 
    GeneratorType gen_type; 
    string generator_args; 
    string sampling_args; 
 
    PredictorImputer() : FeatureProcessor() { init_defaults(); }
 
    // Copy
    //void copy(FeatureProcessor *processor) { *this = *(dynamic_cast<PredictorImputer *>(processor)); }
 
    void init_defaults();
 
    void post_deserialization();
 
    void load_GIBBS(const GibbsSampler<float> &gibbs, const GibbsSamplingParams &sampling_args);
    void load_GAN(const string &gan_path);
    void load_MISSING();
    void load_sampler(unique_ptr<SamplesGenerator<float>> &&generator);
 
    int init(map<string, string>& mapper);
 
    // Learn cleaning model
    int Learn(MedFeatures& features, unordered_set<int>& ids);
 
    // Apply cleaning model
    int _apply(MedFeatures& features, unordered_set<int>& ids);
 
    // Serialization
    ADD_CLASS_NAME(PredictorImputer)
        ADD_SERIALIZATION_FUNCS(processor_type, tag_search, missing_value, gen_type, _sampler,
            generator_args, sampling_args, verbose_apply, impute_features)
};
 
MEDSERIALIZE_SUPPORT(PredictorImputer)
 
 
#endif