MILEAGE PREDICTION USING THREE DIFFERENT MODELS

• I conducted data preprocessing, data visualization, classification tasks and comparison of models to explore and gain a better understanding of different machine learning libraries such as NumPy, Pandas, Scikit-learn, Matplotlib, and Seaborn.

• This project served as a practical learning experience in AIML concepts and tools.

DATASET:

• MPG Dataset – Kaggle Dataset, originally taken from the StatLib library and Maintained at Carnegie Mellon University.

• The dataset was used in the 1983 American Statistical Association Exposition.

• Data was split into a testing set and training test, where 70 percent of data was put in the training test and the remaining 30 percent was put in the testing set.

PERFORMANCE ANALYSIS OF MODELS:

• Initially, the Linear Regression Model was used to make the predictions.

• Later, other models were selected and evaluated - Polynomial Regression and Gradient Regression Models.

• Overall, the best model accuracy observed was in the range of 70% to 77%.

MODELS EVALUATED:

• Linear Regression Model: 70% of the variance in the MPG can be explained by the features in the model.

• Polynomial Regression Model: 75% of the variance in the MPG can be explained by the features in the model.

• Gradient Boosting Regression Model: 77% of the variance in the MPG can be explained by the features in the model.