Maintenance cost prediction for aging residential buildings based on case-based reasoning and genetic algorithm

Building maintenance has been considered a crucial part of the life cycle of a building as the number of aged buildings is increasing globally. A number of maintenance-related problems are known to occur in aged buildings, particularly in urban areas. Thus, great importance should be given to maintenance as it preserves the original function of the buildings and maintains the quality of residents' lives. The degradation incurred by aging of the building can not only lead to excessive repair costs, but also affect its usability and resident safety negatively. Thus, such buildings need to be properly managed using a systematic approach. To this end, it is crucial to predict the maintenance cost for building management as the first step. This research developed a model adopting case-based reasoning and a genetic algorithm to predict the maintenance cost. An experiment was conducted to test the applicability of the model. The predicted maintenance cost was validated with 20 validation datasets. The results demonstrated that the retrieved cases had similarities of approximately 90% on average. The mean absolute error rate of the cost adapted by the Monte Carlo simulation was about 18%, which validated the model. This research contributes to knowledge of building maintenance by not only devising a systematic method for predicting the maintenance cost in advance, but also providing support to building managers’ decisions with regard to maintenance from a long-term perspective. •We devise a model to predict maintenance costs of residential buildings.•The model uses case-based reasoning and a genetic algorithm.•Historical data from a well-known building firm were used to find case similarities.•Monte Carlo simulations lowered the absolute error rate of the model to about 18%.•The model can serve building managers to plan proactive maintenance plans.