Internal Combustion Engine Vehicles Fuel Efficiency Regression Analysis with Physical and Engineering Perspective

Abstract

Internal combustion engine vehicles (ICEVs) have undergone various transformations over the 20th and 21st centuries. A revolutionary in enhancing ICEVs’ fuel efficiency happened particularly around the 1970s under the energy crisis and tightening emission policies. The paper conducted regression analysis on AUTO MPG dataset from UCI Machine Learning Repository to study the underlying relationships between fuel efficiency and various vehicle variables. This helps to make inferences about how adaptive engineering adjustments on those variables enhance fuel efficiency in the 1970s and 1980s under the impact of the energy crisis. The paper introduces physical and engineering connections among given predictor variables from the dataset and derives a general relationship between mpg and its predictors. This substantiates the statistical models’ validity via a rigorous deduced expression. Then, three statistical modelling approaches, namely simple logged factor interaction multilinear regression (MLR), composite-variable MLR, and partial least square (PLS) MLR, were applied for comprehensive analysis and interpretation. Three models have complimentary advantages and drawbacks in interpretability, robustness, and generalizability, but collectively reflect how fuel efficiency evolves and correlated with other variables during 1970 to 1982. Finally, the paper commented on future prospect of ICEV market under the impact of new energy vehicles and explained how ICEVs’ unique advantages made them still a preferable choice in the future.