CHAID algorithm to analyze characteristics of take-out industry

In the restaurant consumption market, with the rapid growth of the take-out industry relying on APP, it is more important to build the take-out key factor model and real-time prediction and recommendation by using the decision tree algorithm of machine learning. In this paper, significant variables were screened by ANOVA through questionnaire survey, and then a decision tree was established by CHAID algorithm. Finally, the three main businesses of the take-out industry are constructed: platform, restaurant and rider feature models. The empirical results show that “platform satisfaction, restaurant satisfaction and rider satisfaction” generate decision trees with 15 nodes at 3 levels, 18 nodes at 2 levels and 17 nodes at 4 levels, respectively. The average correctness is $R = 55.3%, 59.6% and 60.7%. The model prediction accuracy is ideal.