Correlation Analysis Based on Random Forest in Chinese Alcohol Market

摘要

This paper considers the problem of correlation analysis using correlation coefficient and Random Forest. Although Random Forest is a function estimation algorithm, it provides two extra information about independent variables importance according to dependent one and partial dependence. Thus influencing order of correlative factors comes out. The limitation of correlation coefficient of only two variables is broken. An illustration of those skills was done by an application to Chinese alcohol market. The analyzing results indicate that population is the most influencing factor according to liquor. So liquor market at least keep the status quo for slowly population growth.