Clustering of Commercial Vehicles Based on Fuel Type Using Machine Learning Technique

Abstract

This paper delves into the exploration of vehicle characteristics through various clustering algorithms such as Density-Based Spatial Clustering of Applications with Noise, Gaussian Mixture Model, and K-means clustering analysis, focusing primarily on fuel-related attributes. The research problem revolves around uncovering inherent patterns within automobile data to better understand the relationship between different features and fuel types. The methodology involves preprocessing the dataset, which includes fixing missing values errors and encoding categorical variables, followed by scaling the features and applying three clustering algorithms. The Elbow Method is utilized to get the efficient number of clusters, and to find the accuracy of the models, three different evaluation metrics are computed to know the clustering quality. Key findings reveal distinct clusters of vehicles based on fuel-related attributes, providing insights into fuel efficiency and usage patterns across different vehicle models. The analysis is visualized through scatter plots, silhouette scores, the Davies-Bouldin Index, and the Calinski-Harabasz Index with values of 0.528, 0.66, and 57730.245 highlighting the effectiveness of K-means clustering in uncovering meaningful clusters within the dataset. In conclusion, this study demonstrates the utility of clustering analysis in extracting valuable insights from vehicle data, with implications for fuel efficiency optimization and market segmentation in the automotive industry.