The Promotion of Women's Leisure Sports Behavior Based on Improved Decision Tree Algorithm

Introduction

With the improvement of national physical quality and economic development, sports have become an indispensable part of life, and leisure sports have gradually become an important part of people's daily leisure lives (Yu, 2020). Women living in the new era have a great demand for leisure sports and have also spent a lot of energy and financial resources on leisure sports (Zheng, 2018). At present, the research on sports behavior is mostly carried out in cities, considering things such as sports consumption, sports choice, sporting goods, and stadiums, but it is rarely discussed from the perspective of women (Strain et al., 2020).

This paper studies the promotion of women's leisure sports behavior based on an improved decision tree algorithm, which is mainly divided into four sections (Ahn & Chon, 2018). The first section briefly introduces the research background of leisure sports behavior and the arrangement of this study. The second section introduces the research status of behavior at home and abroad and the application and improvement of the decision tree algorithm. This section also summarizes the shortcomings of current research. The third section constructs an analysis model of women's leisure sports behavior based on a decision tree model, improves the decision tree algorithm and the calculation method of information acquisition, and discretizes the continuous attribute problem. In the fourth section, the improved decision tree model constructed in this paper is simulated and analyzed to test the accuracy and running time of the algorithm. Compared with the classical decision tree algorithm, the experimental results show that the improved decision tree algorithm proposed in this paper has advantages in classification accuracy and running time, and has good application value in the analysis of female leisure sports behavior.

The innovation of this paper is the improvement strategy of decision tree analysis using the Taylor formula and McLaughlin expansion formula to improve the information gain rate of the C4.5 approximation algorithm, simplify the algorithm, improve the accuracy of the algorithm, and avoid logarithmic operation. In addition, aiming at the problem of discretization of continuous attributes, the discretization method based on the chi-square value is used to obtain the alternative optimal splitting breakpoint value to ensure classification accuracy.