Predicting gas usage as a function of driving behavior

The driving behavior, road, and traffic conditions greatly affect the gasoline consumption of an automobile. We believe that accurate prediction of gasoline usage as a function of the above can be used to obtain optimal driving behavior, for given road and traffic conditions, to reduce gasoline consumption. This results in great dollar savings and also reduces the harmful effects of gasoline usage on environment. In this project, we propose machine learning framework to predict gasoline usage as a function of driving behavior with high accuracy. Analysis of the results shows that we are able to predict the rate of gasoline consumption (Miles/hour) much better than fuel economy (Miles/gallon), as a function of driving behavior. We then propose an application by constructing a machine learning framework to minimize the total gasoline consumption over a period of time.