Predicting Pavement Condition Index Using Machine Learning Algorithms and Conventional Techniques

Government agencies and transportation engineers use pavement management systems (PMS) to evaluate performance keep above the minimum acceptable standards. The Pavement Condition Index (PCI) international roughness index (IRI) are among most commonly used indices conditions. Due IRI data collection being more accessible less expensive than collecting distress data, this study aims develop PCI models that can successfully estimate values based on for flexible using two Machine Learning techniques (ML), namely: Random Forest (RF), Support Vector (SVM), three conventional techniques, linear, quadratic, cubic regression. was carried out with database collected from Long-Term Performance (LTPP) program. results of dataset reveal both ML (RF SVM) have strong prediction ability high coefficient determination (R^2 = 99.7 96.8) %, low Root Mean Squared Error (RMSE 1.095 3.569) % Absolute (MAE 0.474 2.244). In conclusion, goodness fit proposed compared previously developed. showed yielded higher accuracy techniques.