timeseriessplit.h

Go to the documentation of this file.

//
// Created by Xinyu Zhang on 4/6/21.
//
 
#ifndef COSAN_TIMESERIESSPLIT_H
#define COSAN_TIMESERIESSPLIT_H
 
#include <cosan/selection/selection.h>
namespace Cosan{
    class TimeSeriesSplit: public Splitter {
        public:
            TimeSeriesSplit() : Splitter() {}
            TimeSeriesSplit(gsl::index kfoldnumber) : Splitter(kfoldnumber) {}
            TimeSeriesSplit(gsl::index nrows, gsl::index kfoldnumber): Splitter(nrows,kfoldnumber){}
            void SetSplit(gsl::index nrows){
                if (nrows <= KFoldNumber) {
                    throw SmallRows;
                }
                fmt::print("*********************************\n");
                fmt::print("Begin Time Series Splitting!!\n");
 
                KFoldNumber= KFoldNumber;
                std::vector <gsl::index> idx(nrows);
                std::iota(idx.begin(), idx.end(), 0);
                gsl::index foldSize = nrows/(KFoldNumber+1);
                for (gsl::index i = 1; i<=KFoldNumber;i++){
//                    std::vector<gsl::index> trainidx = std::vector<gsl::index>(idx.begin(), idx.begin()+foldSize*i+nrows%(kfoldnumber+1));
//                    std::vector<gsl::index> testidx  = std::vector<gsl::index>(idx.begin()+foldSize*i+nrows%(kfoldnumber+1), idx.begin()+foldSize*(1+i)+nrows%(kfoldnumber+1));
//                    split_batch.push_back({trainidx,testidx});
                    split_batch.push_back({ std::vector<gsl::index>(idx.begin(), idx.begin()+foldSize*i+nrows%(KFoldNumber+1)),
                                            std::vector<gsl::index>(idx.begin()+foldSize*i+nrows%(KFoldNumber+1), idx.begin()+foldSize*(1+i)+nrows%(KFoldNumber+1))});
                    fmt::print("Current Index is {:}, trainidx size:{:}, testidx size:{:}\n",
                               i-1,foldSize*i+nrows%(KFoldNumber+1),foldSize);
                }
                fmt::print("End Time Series Splitting!!\n");
                fmt::print("*********************************\n");
            }
 
           std::vector< std::tuple<std::vector<gsl::index>,std::vector<gsl::index> > > GetSplit() {return split_batch;}
        private:
            std::vector< std::tuple<std::vector<gsl::index>,std::vector<gsl::index> > > split_batch;
 
    };
 
    class TimeSeriesSplitParallel: public Splitter {
    public:
 
        TimeSeriesSplitParallel() : Splitter() {}
        TimeSeriesSplitParallel(gsl::index kfoldnumber) : Splitter(kfoldnumber) {}
        TimeSeriesSplitParallel(gsl::index nrows, gsl::index kfoldnumber): Splitter(nrows,kfoldnumber){}
 
 
        void SetSplit(gsl::index nrows){
            if (nrows <= KFoldNumber) {
                throw SmallRows;
            }
            std::vector <gsl::index> idx(nrows);
            std::iota(idx.begin(), idx.end(), 0);
            gsl::index foldSize = nrows/(KFoldNumber+1);
//            std::mutex mylock;
            fmt::print("*********************************\n");
            fmt::print("Begin Time Series Splitting in Parallel!!\n");
            split_batch.resize(KFoldNumber);
            #pragma omp parallel for
            for (gsl::index i = 1; i<=KFoldNumber;i++){
//                    std::vector<gsl::index> trainidx = std::vector<gsl::index>(idx.begin(), idx.begin()+foldSize*i+nrows%(kfoldnumber+1));
//                    std::vector<gsl::index> testidx  = std::vector<gsl::index>(idx.begin()+foldSize*i+nrows%(kfoldnumber+1), idx.begin()+foldSize*(1+i)+nrows%(kfoldnumber+1));
//                    split_batch.push_back({trainidx,testidx});
//                mylock.lock();
                fmt::print("Current Index is {:}, the current thread num is {:}, total number of threads {:}. trainidx size:{:}, testidx size:{:}\n",
                           i-1, omp_get_thread_num(),omp_get_num_threads(),foldSize*i+nrows%(KFoldNumber+1),foldSize);
                split_batch[i-1]={ std::vector<gsl::index>(idx.begin(), idx.begin()+foldSize*i+nrows%(KFoldNumber+1)),
                                   std::vector<gsl::index>(idx.begin()+foldSize*i+nrows%(KFoldNumber+1), idx.begin()+foldSize*(1+i)+nrows%(KFoldNumber+1))};
//                mylock.unlock();
            }
 
            fmt::print("End Time Series Splitting in Parallel!!\n");
            fmt::print("*********************************\n");
        }
        std::vector< std::tuple<std::vector<gsl::index>,std::vector<gsl::index> > > &GetSplit() {return split_batch;}
    private:
        std::vector< std::tuple<std::vector<gsl::index>,std::vector<gsl::index> > > split_batch;
    };
}
 
#endif //COSAN_TIMESERIESSPLIT_H