Start Date
6-1-2011 1:00 PM
End Date
6-1-2011 2:15 PM
Track
1. Track 1 – Formal Paper Presentation
Subject Area
Food Service
Faculty Member
Lisa Slevitch, lisa.slevitch@okstate.edu
Abstract
Within the competitive foodservice industry, the ability to accurately predict the length of the meal process known as turn-time is critical to the success of the firms in the industry. This is traditionally done through multiple least squares (linear regression) technique. However, linear regression lack the characteristics needed to accurately predict time durations, while survival models were designed for that purpose. This study utilized simulated data of a dine-in restaurant to test and compare the ability of linear regression to five survival models (proportional hazard models) to accurately predict the duration of turn-time. The results from the simulated trials show that while some of the survival models held marginal improvements, linear regression performed adequately for predicting duration of turn-time as compared to the survival models. For practitioners interested in the practical ease of the models, linear regression is recommended while practitioners interested in incremental improvements may opt for survival models.
Keywords
Survival modeling, proportional hazards models, turn-time, revenue management, simulations
Using Survival Modeling for Turn-Time Predictions in Foodservice Settings
Within the competitive foodservice industry, the ability to accurately predict the length of the meal process known as turn-time is critical to the success of the firms in the industry. This is traditionally done through multiple least squares (linear regression) technique. However, linear regression lack the characteristics needed to accurately predict time durations, while survival models were designed for that purpose. This study utilized simulated data of a dine-in restaurant to test and compare the ability of linear regression to five survival models (proportional hazard models) to accurately predict the duration of turn-time. The results from the simulated trials show that while some of the survival models held marginal improvements, linear regression performed adequately for predicting duration of turn-time as compared to the survival models. For practitioners interested in the practical ease of the models, linear regression is recommended while practitioners interested in incremental improvements may opt for survival models.