TITLE:
Application of Decision Tree Algorithm in Housing Purchase Problems—A Case Study of Xining City
AUTHORS:
Siyu Chen, Li Fu
KEYWORDS:
Decision Tree, Gini Coefficient, Correlation Coefficient
JOURNAL NAME:
Journal of Computer and Communications,
Vol.12 No.11,
November
27,
2024
ABSTRACT: Decision tree is an effective supervised learning method for solving classification and regression problems. This article combines the Pearson correlation coefficient with the CART decision tree, replacing the Gini coefficient with the correlation coefficient to consider the correlation between conditional attributes, prioritizing the selection of conditional attributes with higher correlation coefficients as leaf nodes. The collected data on homebuyers is divided into age groups, including youth, middle-aged, and elderly groups. Both traditional CART decision tree and improved CART decision tree are applied to this problem, and after comparison, it is found that the depth of the CART decision tree in this study is reduced, the number of leaf nodes is decreased, the time complexity is shortened, efficiency is improved, and pruning issues are avoided. Finally, corresponding housing recommendations are given to homebuyers of different ages.