基于数据挖掘的川菜食谱数据多维度可视分析  

Multi-dimensional Visual Analysis of Sichuan Cuisine Recipe Data Based on Data Mining

作　　者：李苗苗 华才健 赵长青 LI Miaomiao;HUA Caijian;ZHAO Changqing(School of Computer Science and Engineering,Sichuan University of Science&Engineering,Yibin 644000,China;School of Biological Engineering,Sichuan University of Science&Engineering,Yibin 644000,China)

机构地区：[1]四川轻化工大学计算机科学与工程学院,四川宜宾644000 [2]四川轻化工大学生物工程学院,四川宜宾644000

出　　处：《四川轻化工大学学报(自然科学版)》2025年第1期77-85,共9页Journal of Sichuan University of Science & Engineering(Natural Science Edition)

基　　金：四川省科技厅项目(2019YFG0169);四川轻化工大学研究生创新基金项目(Y2023118)。

摘　　要：针对川菜研究较少关注味型、烹调技法及耗时等多维度数据分析,提出了一种基于数据挖掘的川菜食谱数据多维度可视分析方法。首先,通过旭日图表达川菜多维属性,分析川菜构成特征;其次,对川菜食谱进行聚类分析,通过优化初始聚类中心选取,得到的方差比准则指数和轮廓系数均最大(最大值分别为3.657和0.359),聚类效果更优,解决了传统K-means算法聚类结果不稳定问题;然后,采用关联规则分析食谱成分的惯用搭配;最后,通过TF-IDF算法对菜品进行加权,并计算余弦相似度。分析结果表明,川菜中脂肪含量较高,主料中动物性原料占比71.00%,味型以辣味和咸鲜味最为显著,共享食材越多则相似度越高。所提出的方法能够从多维度分析川菜的构成特征及食谱间的关联,帮助用户发现川菜食谱内部的特征和联系。In view of the fact that the research on Sichuan cuisine pays less attention to multi-dimensional data analysis such as flavor,cooking techniques and time-consuming,a multi-dimensional visual analysis method of Sichuan cuisine recipe data based on data mining is proposed.Firstly,the sunburst map is used to express the multi-dimensional attributes of Sichuan cuisine,and the composition characteristics of Sichuan cuisine are analyzed.Secondly,the clustering analysis of Sichuan cuisine recipes is carried out,and by optimizing the selection of the initial clustering centers,the variance ratio criterion index and contour coefficient are obtained to be the largest(the largest volues are 3.657 and 0.359,respectively),and the clustering effect is better,which solves the problem of unstable clustering results of the traditional K-means algorithm.Then,association rules is used to analyze the common collocation of recipe ingredients.Finally,the dishes are weighted by the TF-IDF algorithm and the cosine similarity is calculated.The analysis results show that the fat content of Sichuan cuisine is high,the proportion of animal raw materials in the main ingredients is 71.00%,the spicy and salty umami are the most significant flavor types,and the more shared ingredients,the higher the similarity.The proposed method can analyze the composition characteristics of Sichuan cuisine and the association between recipes from multiple dimensions,helping users to discover the internal characteristics and associations of Sichuan cuisine recipes.

关 键 词：数据挖掘 多维数据可视化 K-MEANS算法 关联规则 TF-IDF算法 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]