BinSplitOptimizer.h

#ifndef __BIN_SPLITTER_H__
#define __BIN_SPLITTER_H__
#include <map>
#include <vector>
#include <limits.h>
#include <SerializableObject/SerializableObject/SerializableObject.h>
 
using namespace std;
 
class BinSplitOptimizer {
public:
    vector<int> SplitToBins(const vector<float> &vec, const vector<float> &y, size_t kBins,
        size_t min_samples, vector<float> &histAvg, vector<float> &partitionValues, bool &has_error);
private:
    bool evaluate(const vector<int> &indexes, float &totScore);
 
    float *_sortedArr;
    int _minSamples;
    map<float, int> _histElements;
    map<float, float> _histAvg;
 
};
 
enum BinSplitMethod {
    SameValueWidth, 
    PartitaionMover, 
    IterativeMerge, 
    DynamicSplit, 
};
 
class BinSettings : public SerializableObject {
public:
    int min_bin_count; 
    double min_res_value; 
    int binCnt; 
    float min_value_cutoff; 
    float max_value_cutoff; 
    bool weighted; 
 
    BinSplitMethod split_method; 
 
    int init(map<string, string>& map);
 
    static const unordered_map<int, string> name_to_method;
    static BinSplitMethod bin_method_name_to_type(const string& bin_method);
 
    BinSettings() {
        min_bin_count = 1;
        min_res_value = 0;
        binCnt = 10;
        min_value_cutoff = (float)INT_MIN;
        max_value_cutoff = (float)INT_MAX;
        weighted = false;
    }
 
    ADD_CLASS_NAME(BinSettings)
    ADD_SERIALIZATION_FUNCS(min_bin_count, min_res_value, binCnt, min_value_cutoff, max_value_cutoff, 
        split_method, weighted)
};
 
MEDSERIALIZE_SUPPORT(BinSettings)
namespace medial {
    namespace process {
        void split_feature_to_bins(const BinSettings &setting, vector<float> &feature,
            const vector<int> &sel_indexes, vector<float> &y);
        void normalize_feature_to_uniform(const BinSettings &setting, vector<float> &feature);
    }
}
 
#endif