1. ¹McGill University, Montreal, QC, Canada
2. ²Clemson University, Clemson, SC, USA
3. ³University of Pittsburgh, Pittsburgh, PA, USA

Correspondence: KJ Glowacka, McGill University, Desautels Faculty of Management, Montreal, QC, Canada H3A 2T5

Received November 2007; Accepted November 2008; Published online 11 March 2009.

Abstract

This paper considers the outpatient no-show problem faced by a rural free clinic located in the south-eastern United States. Using data mining and simulation techniques, we develop sequencing schemes for patients, in order to optimize a combination of performance measures used at the clinic. We utilize association rule mining (ARM) to build a model for predicting patient no-shows; and then use a set covering optimization method to derive three manageable sets of rules for patient sequencing. Simulation is used to determine the optimal number of patients and to evaluate the models. The ARM technique presented here results in significant improvements over models that do not employ rules, supporting the conjecture that, when dealing with noisy data such as in an outpatient clinic, extracting partial patterns, as is done by ARM, can be of significant value for simulation modelling.