TPJF: Machine Learning Based Intelligent Prediction of Preference for Japanese Food
Peiran Yu, Min Fu · 2020
With the development of the catering industry in China in recent years, foreign cuisine, such as Japanese cuisine, has become a significant part of some people's daily lives. However, for Japanese restaurants, there has not been a widespread and reliable way to target at advertising to those interested in Japanese food, or for consumers to determine whether they are interested in Japanese food. This paper proposes an intelligent and novel prediction method, called TPJF – Targeting Preference for Japanese Food, to determine people's attitude towards Japanese food. Our method is proposed based on the dataset obtained by real world questionnaires, which are about participant fillers’ basic dietary habits and personal information. We apply data mining techniques on the dataset. During the evaluation experiments, machine learning models such as DT, KNN, GBM, RF, ANN, SVM and NB are investigated to achieve such a multi-class classification. The experimental results show that RF offers the best accuracy of 76.26% and the best F1-score of 77.67%. RF also provides the optimal effective accuracy of 74.29% for advertising, which is defined by this paper.