Prediction of maintenance cost for road construction equipment: a case study

Equipment maintenance cost is significant in construction operations budgets. This study proposes a systematic approach to predict maintenance cost of road construction equipment. First, maintenance cost data over more than 10 years was collected from a partner company's equipment management information system. Data was cleaned and analyzed to obtain a general understanding of maintenance costs trends. Next, traditional cumulative cost models and alternative data mining models were generated to predict maintenance cost based on available equipment and operation attributes. Data mining models were evaluated and validated using portions of the collected data that have not been used in model development. Data collection, analyses, modeling, and validation steps are discussed. The paper also presents the performance of different model types. Based on the case study data, regression model trees performed better than other model types with equipment work hours being the most significant parameter for predicting maintenance cost.