A Time Series Case-Based Predicting Model for Reservation Forecasting

摘要

This study addresses how to construct sales forecasting models by using restaurant reservation data. The issues of how to retrieve booking patterns, search for influential parameters, and divide samples for training, validating, and testing are discussed. Regression and Pick Up models, which are common practice, are also built as benchmarks. We used data from a mid-sized restaurant to show that the proposed Time Series Case-Based Predicting model can significantly outperform the benchmarks in all testing cases.