A Time Series Regression Model via Improved PCA and Bagging Algorithms

Considering the case that prediction variable is a time series and response continuous scalar, we propose regression model based on improved PCA Bagging Algorithms. Compared with dimension reduction, proposed method uses distance correlation coefficient matrix instead of Person matrix, which makes distribution assumption original variables more free. an unsupervised reduction technique connection functions between principal components are unknown, to use Algorithmss capture information related variables. In actual data analysis, comparative methods LASSO PCA-based linear models, empirical results show has certain competitiveness compared comparison method.Finally, because base-model Algorithms model-free, some machine learning higher precision flexibility can be used as for tasks different complexity.