Machine Learning Predicts Truck Breakdowns in Indonesia with 83% Accuracy

Abstract

PT. Varia Usaha Beton, a cement product company, faces frequent breakdowns of mixer trucks, reducing reliability from the target 90% to 60%. This study aims to predict truck breakdowns using a machine learning model based on the K-NN algorithm within the CRISP-DM framework. Data from the company's maintenance records were cleaned and split into training and testing sets. With k=20, the model achieved 90% accuracy on training data and 83% on testing data. These results can help improve maintenance scheduling and resource planning, enhancing truck reliability. Future research should compare other algorithms and consider different programming environments. Highlights:   High Accuracy: K-NN model achieved 90% training and 83% testing accuracy. Maintenance Aid: Improves scheduling and resource planning for truck maintenance. Future Research: Compare algorithms and explore different programming environments.   Keywords: Predictive Maintenance, Mixer Trucks, K-NN Algorithm, CRISP-DM, Machine Learning