Attribute level analysis on food product choice using latent class choice model

Abstract

With the rapid growth of Home Meal Replacement (HMR) market recently, ready-to-cook kit home delivery service has emerged as a new business model. The service delivers pre-portioned ingredients for customers to cook whom either do not have time to spend time on grocery shopping or do not have the effort. As far as I know, attempts to examine people’s preference on food products at the attribute level has been minimal. Unlike goods that have been traditionally used in preference measurement, food products are both functional and hedonic in nature where consumer heterogeneity exists not only in the importance weights of the attributes but also how attribute levels are perceived. In this thesis, I suggest a methodology that provides a structure for representing food products with attributes, namely taste and cooking methodology. To empirically validate this method, we apply the method to a dataset of consumer purchases for a ready-to-cook kit delivery service over four weeks. The latent class choice model is adopted for the analysis since it is expected that there may be multiple segments in the preference structure of the customers group that is examined. The result of the analysis shows that there are three classes of customer with different preference over the menus. Two classes showed distinct preference over the cooking methods, while one class is shown to have no particular preference over the product features. The analysis could further be applied to investigate consumer preferences for food products in a field experiment setting by examining sales increase for target segments.