基于决策树算法的青年女性腰臀部识别

Waist and Hip Shape Recognition of Young Women Based on Decision Tree Algorithm

为提高青年女性腰臀部分类的准确性,为个性化定制提供依据,构建了一种基于决策树算法的青年女性腰臀部识别模型。采用聚类分析对收集到的153名18~25岁的青年女性腰臀部数据进行分类,提取5项反映腰臀部形态的典型指标,根据指标将腰臀部形态分为3类;再运用CART决策树算法将样本数据按70∶30的比例划分为训练集和测试集,并根据基尼系数选择最佳节点和最佳分支,构造分类树,研究青年女性腰臀部形态的分类规则,以可视化的形式来展现结果。结果表明:青年女性的臀部形态可分为3种,即凸臀体、正常体和凸腹体。决策树算法对腰臀部的分类识别效果良好,是一种切实有效的算法。

In order to improve the accuracy of classifications of young women's waist and hip,and provide the basis for personalized customization,a young women's waist and hip recognition model based on decision tree algorithm was constructed.Cluster analysis was used to classify the waist and hip data of 153 young women aged from 18 to 25,and five typical indices reflecting the waist and hip morphology were extracted.According to the indices,the waist and hip morphology was divided into three categories.Then the cart decision tree algorithm was used to divide the sample data into training set and testing set according to the proportion of 70:30,and the best node and branch was selected according to the Gini coefficient.It constructed the classification tree,and studied the classification rules of young women's waist and hip shape,and showed the results in a visual form.The results show that the hip shape of young women can be divided into three types including convex hip body,normal type and convex abdomen body.The decision tree algorithm has a good effect on the classification and recognition of waist and hip,which proves to be a practical and effective algorithm.