import binary_optimization as opt#import library
import numpy as np
from sklearn import svm

class Evaluate:#setting class
    def __init__(self):#set train_data,train_label
        self.tr_x = tr_x
        self.tr_y = tr_y
        self.K = 4
    def evaluate(self,gen):
        """
        Setting of evaluation function.
        Here, the correct answer rate is used.
          anser_label/all_label
        """
        mask = np.array(gen) > 0
        al_data = tr_x[:,np.nonzero(mask)[0]]
        kf = ms.KFold(n_split=self.K, shuffle=True);s=0
        for tr_ix, te_ix in kf.split(self.tr_x):          
          s += svm.LinearSVC().fit(al_data[tr_ix],self.tr_y[tr_ix]).score(al_data[te_ix], self.tr_y[te_ix])
        return s/self.K
        #↑evaluate with fittness function
    def check_dimentions(self,dim):#check number of all feature
        if dim==None:
            return len(self.train_d[0])
        else:
            return dim

print("Algorithm:\n\t{0}  {1} {2}".format("best_pos","best_val","number_of_1s"))

s,g,l=opt.BGA(Eval_Func=Evaluate, n=20, m_i=200)#score, gen_list, gen length of 1
print("BGA:\n\t{0}   {1}  {2}".format("".join(map(str,g)),s,l))