Predictive maintenance is an early-diagnosis-based approach for upholding machine health by resolving potential problems in advance. It has a significant role in the automotive sector due to its cost savings. Predictive maintenance for automotive vehicles allows users to identify whether an automotive vehicle needs maintenance at a given instance. Generally, abnormalities in various parameters associated with these vehicles are used for this purpose, while the methodology for detecting such abnormalities may vary. This approach also helps predict the potential lifespan of these vehicles. Machine learning is one of the most crucial methods that can help achieve this. A machine learning-based system for the early prediction of the need for maintenance of automotive vehicles is presented in this paper. Models accurately learn various trends and relations between these parameters and automotive health. The trained and tested models include Random Forest, Decision Trees, K-Nearest Neighbors, Naïve Bayes, and others on the Engine Health dataset. A preprocessing pipeline, precisely curated for this dataset, is also presented. A novel accuracy of 76% was obtained, which is significantly higher than all the previously proposed methods on the Engine Health dataset. Finally, a prototype and some practical deployment methodologies for the proposed models are discussed as a proof of concept.