Developing a Strategy for Vehicle Pollution Reduction

Background: Carbon Dioxide (CO2) is a greenhouse gas that constitutes only one percentage of atmospheric gases but is responsible for about eighty one percentage of total emissions. CO2 is naturally produced through decomposition, respiration, and ocean processes. However, human endeavours like the combustion of fossil fuels, cement production, and automotive exhaust also release CO2 .Aims: This paper aims to predict CO2 emissions from automobiles in real-time. We focus to predict CO2 emissions and perform a comparative study to achieve higher efficiency .Methods: In order to predict the response variable, multiple linear regression requires more than one predictor variable, making it an extension of simple linear regression.. A regression line is attempted to be fit by MLR via a multidimensional space of data points. Multiple linear regression expresses Y which is the target as a linear composition of the predictor variables. Support Vector Machines (SVM), known for their robustness, are effective in performing regression, identifying outliers, and classifying data in both linear and nonlinear manners. Results: The study's analysis of the data leads to the conclusion that the random forest (RF) and support vector machine (SVM) algorithms for machine learning, successfully predicted CO2 emissions and yielded better results .Discussion: The study also suggests that these two algorithms are the most suitable for CO2 emission prediction. Based on the analysis from the same study, the ML algorithms proposed here were compared using the RMSE and R2 parameters to determine which algorithm performed better in predicting CO2 emissions. Conclusions: On considering both the parameters and on analyzing them Multiple Linear regression produced a better result than Support Vector Machine algorithm (SVM)