Outcome Prediction of Construction Change Disputes using Machine Learning

Going through a Construction Dispute is one of the most stressful events that might take place throughout the course of a project. the time it takes to resolve a Dispute in North America is (14-18) months. If the parties to a construction dispute can forecast the conclusion of the case with some degree of certainty (prior to proceeding with the actual dispute), they will have a greater likelihood of settling the case outside of court even faster. Many researchers studied the use of machine Learning algorithms in predicting the outcome of disputes. They were able to achieve working models up to 91.15% accuracy. This study is focusing on [1] increase the prediction accuracy by limiting the number of variables used in making the prediction. [2] utilized 4 different algorithms (Decision Tree, Random Forest, Neural Network, SVM) to choose the one with better performance The results showed that Random Forest model was the better performing model which achieved (95.0%) accuracy, and outperformed the models developed by former studies.