A univariate feature selection approach for finding key factors of restaurant business

Abstract

With the help of globalization, fast food has become very popular in Bangladesh as it is concerned with the taste and habit of the people. Based on the customers' choice, there are many factors associated with it such as rating, reviews, environment, publicity, and so on. Hence, it is vital to evaluate customers' opinion to observe the reasons behind their preference. The objective of this research is to get the insight of the young generations' fast food preference with respect to relevant key features in order to gain restaurant business success. 170 respondents were gathered based on a structured questionnaire to conduct the research. Attributes were selected using the univariate feature selection method. Selected features were then used in supervised machine learning models. Gaussian Naïve Bayes, decision tree classifier (CART), random forest classifier and logistic regression were used to predict students' fast food consumption rate. Among these machine learning classification techniques, Naive Bayes performed best with 79.4% accuracy by correctly classifying the highest number of instances. The result concludes university students' preference associated with restaurant factors and finds potential insights to fast food restaurant business.