Random Forest Regression Analysis for the Failure Prediction of Corrosion in Pipeline Infrastructure

Pipeline failure prediction is crucial in ensuring optimal pipeline maintenance strategies. This work applies a Random Forest Regression model (RF) for failure prediction of pipeline using wall of thickness and depth of corrosion. It highlights the data cleaning procedures applied on the Inline Inspection Dataset (ILI), dimensionality reduction using Principal Component Analysis (PCA) and evaluate the performance of different Random Forest Regression models using the coefficient of determination (R
²
). Finally, a recommendation for further research is proposed.