The idea for this project came about when all members of the team realized that we might have paid more than we should’ve for our recently acquired used cars. With an aim to come up with a method to better predict the value of a used car, we modeled the problem as as regression problem and implement common regression models such as Linear Regression and Random Forests as well state of the art models and frameworks such as XGBoost and LightGBM.
Our best models achieved R2 scores of about 0.88 after data pre-processing and training.