Research on PCA Data Dimension Reduction Algorithm Based on Entropy Weight Method

Traditional principal component analysis (PCA) is a common linear dimension reduction algorithm, but its dimension reduction effect is relatively poor, and the algorithm takes a long time, and it can not meet the prediction target well when applied to the actual scene. Therefore, by introducing the concept of mutual information in information theory and combining with the idea of entropy weight method, an improved PCA algorithm ew-pca is proposed. Firstly, mutual information threshold is set for feature screening, and then the concept of weighted average value is proposed to improve the data centralization process. Finally, the entropy weight is introduced to improve the principal component to optimize the dimensionality reduction process. KNN and SVM algorithm are used to predict and analyze the processed data set. Compared with the traditional PCA algorithm, the improved ew-pca algorithm has better dimension reduction effect and higher prediction accuracy.