Development of initial costs forecasting models for school buildings using multiple linear regression

The recent economic crises affected the whole world economic which enforced some countries to declare their bankruptcy such as Argentina and Island. Government of Canada spends yearly billions of dollars to construct and run school buildings which represent the major domain and largest footprint of public sector. Since that the initial costs proved to have the highest impact of the life cycle costs in school buildings in Canada, and since that these initial costs are affected by structure and envelope types of school, Linear regression models are developed to enable school boards to predict the initial costs of new conventional school buildings associated with applying various structure and envelope types. Seven multi-regression models are developed in this paper to predict the square footage initial costs for different structure and envelope types, covering steel, concrete, and wood structures, in various combinations for new conventional school buildings. The RS Means is used in this study to predict the construction costs and to create 420 data points to be used in the models. Each model is developed to predict the specified structure and envelope type with regards to correlated predictor variables that include: school area (square foot), number of floors (which ranges from 1-4), and school level, which ranges from 1-3: elementary school (1), middle school (2), and high school (3). The development of the models is performed throughout three major stages; preliminary diagnostics on data quality, the models development process, and models validation. The developed models will help the school boards to predict the initial costs and to select the economical alternative.