Affiliation:
1. Marathwada Mitra Mandal’s College of Engineering, Pune, Maharashtra, India
Abstract
The traditional way to calculate fuel economy is done by using odometer reading and fuel consumed by car to travel that particular distance. This is a very narrow approach as fuel economy is affected by a variety of factors in the real world. Features such as throttle response, engine temperature, coolant temperature, gross weight of vehicle, etc. have a huge influence on the fuel economy. In order to overcome this problem, we have tried to predict fuel economy based on various features extracted from telemetric data in our project. In order to achieve this, we have implemented various feature selection and feature extraction techniques by further analyzing them with the purpose of calculating the effectiveness of those features to achieve high performance of machine learning algorithms that ultimately improves the predictive accuracy of the classifier. This provides us with the information regarding the amount of influence a particular feature has on the overall fuel economy of the vehicle.