20201210 이승훈
I tried as many models as possible and in this dataset, it turns out linear(logistic) and SVM, MLP works better than other models like decision tree, random forest etc. Thus I find proper parameters for several SVM models and linear models, and some other models, and finally I choose to use soft voting and hard voting to increase overall accuracy. Since the nunber of models are about 7, accuracy increases well.
The final accuracy is about 99.17%
#MIT licnese