Recently, many construction organizations have shifted from offline to online data storage platforms, such as MS SharePoint. Construction companies are storing more historical data than ever. The question is how these historical data can be structured and analyzed to make actionable decisions and bring competitive advantages. This dissertation discusses a reliable solution for using historical data in construction. In this project, we developed a decision support system which predicts conceptual costs of construction projects and supports decision-making for long-term capital planning in public universities. A prototype system was developed based on historical data for roofing projects at the University of Alabama. We collected this historical data via a web-based data entry form subsystem. The developed system uses ridge regression models to train historical data, which helps reduce multicollinearity. K-fold cross-validation and evolutionary algorithm are used to fit ridge regression models. This system has a user-friendly interface and supports what-if analysis which allows the user to see multiple scenarios of the estimation. Also, the system has an automated process which adjusts inflation based on construction start date. Finally, validation has shown that the system can improve in the accuracy in conceptual estimation of roofing projects at the University of Alabama.