RMM: An R Package for Customer Choice-Based Revenue Management Models for Sales Transaction Data

TL;DR

This paper introduces an R package called RMM that implements a customer choice-based revenue management model using the Conditional Logit model and Robust Demand Estimation, facilitating demand estimation from hotel transaction data with frequent price changes.

Contribution

The paper presents a new R package RMM that applies the CL model with RDE to improve demand estimation from transaction data without data aggregation.

Findings

Successfully applied to hotel transaction data

Provides estimates of choice probabilities and no-purchase customer size

Enables demand modeling with frequent price changes

Abstract

We develop an R package RMM to implement a Conditional Logit (CL) model using the Robust Demand Estimation (RDE) method introduced in Cho et al. (2020), a customer choice-based $\textbf{R}$evenue $\textbf{M}$anagement $\textbf{M}$odel. In business, it is important to understand customers' choice behavior and preferences when the product prices change over time and across various customers. However, it is difficult to estimate demand because of unobservable no-purchase customers (i.e., truncated demand issue). The CL model fitted using the RDE method, enables a more general utility model with frequent product price changes. It does not require the aggregation of sales data into time windows to capture each customer's choice behavior. This study uses real hotel transaction data to introduce the R package RMM to provide functions that enable users to fit the CL model using the RDE method along with estimates of choice probabilities, size of no-purchase customers, and their standard errors.