ROAD CONSTRUCTION COST ALLOCATION AND FORECASTING USING MACHINE LEARNING TECHNIQUES

Abstract

Cost forecasting and allocation is the most widely used method for managing, planning, and addressing the objectives of goals. Ineffective cost allocation and forecasting are posing a big challenge in road construction projects. Underestimation is one of the major challenges in the construction phase, which leads to projects are not accomplished in desired scope, quality, time, and budget. Therefore, the proposed study aimed to develop a road construction cost allocation and forecasting model using supervised machine learning algorithms such as support vector regression, random forest regression, boosting regression, k-nearest neighbor regression, and decision tree regression to improve the performance of existing cost allocation and forecasting of road construction projects in Ethiopia with special reference to Wolayita city municipality. Well organized and optimal size of the dataset collected from the Wolayita city municipal organization, which includes a five-year road construction cost recorded data from the years 2008–2013 E.C. a total of 10,158 instances with 17 features preprocessed and 9000 instances with 14 essential features with target variable are used to develop the proposed model. Best-fit machine learning algorithms are necessary to develop proposed road construction projects' cost allocation and forecasting models. The researchers used a Python tool to analyze and evaluate the proposed model, which was built on a percentage split of 80/20 training and testing of the selected instances of the dataset. Regression performance metrics are used to compare and identify the best model to develop for the cost allocation and forecasting models of road construction projects. The main findings of this study are outcomes random forest regression algorithm with an accuracy of 98.8% in the training set and 98.5% in the testing set and the developing a smart, simplified, and efficient cost allocation and forecasting model for road construction projects in Ethiopia with the system prototype and tested the cost allocation and forecasting ability