Virtualized Deployment of Predictive Analytics in the Automotive Sector: Auto-scaling for containerized applications

This paper focuses on developing a predictive analytics application tailored for the automotive industry. The primary aim is to harness vast quantities of data from vehicle operations and consumer behaviors to enhance strategic decision-making and market competitiveness. Utilizing the ’quikr-cars-scraped’ dataset, the application employs advanced machine learning models to analyze vehicle specifications and market trends, providing actionable insights into car pricing, carbon emissions, launch year, and usage patterns. The proposed architecture is built for scalability and easy deployment, leveraging containerization with Docker to ensure consistent performance across various computing environments. A multi-model machine learning pipeline is integrated, featuring algorithms such as Linear Regression, Decision Trees, Random Forest, and Support Vector Machines, each chosen for its efficacy in handling specific types of predictive tasks. The Flask web application serves as the user interface, allowing real-time data interaction and retrieval of predictions, facilitating an intuitive user experience. The implementation involves preprocessing data to ensure accuracy, developing and testing individual predictive models, and deploying these models in a structured pipeline within the Flask application. The final product demonstrates significant potential to transform automotive data into strategic assets, driving innovations and operational efficiencies in the automotive sector.