Small and medium-sized enterprises (SMEs), especially those in the restaurant sector, often lack the resources and expertise needed for data-driven decision making, which puts them at a competitive disadvantage. This paper addresses this gap by proposing a practical two-stage machine-learning framework to predict when restaurant customers are likely to report high satisfaction. Using a publicly available dataset of 1,500 customer records, the first stage uses a Chi-square test to identify significant, actionable factors, such as service rating, online reservation and meal type, while removing demographic variables that do not contribute useful information. The second stage applies logistic regression to classify and predict high satisfaction. Because the dataset is heavily imbalanced (far fewer ‘high satisfaction’ cases), the study shows that undersampling is necessary, with random undersampling producing the best balance and an AUC—ROC of 0.833. The results give SME restaurant managers a clear set of evidence-based priorities for improving customer experience and a practical tool for forecasting how customers may respond to future changes, helping reduce operational risk and improve marketing efficiency. This article is also included in The Business & Management Collection which can be accessed at http://hstalks.com/business/.
Dinh et al. (Sun,) studied this question.