Multiple Linear Regression Algorithm-based Car Price Prediction

Unlike "real estate", cars are depreciating assets, which means that their value over time and with use, decreases; because of this very reason, people find it very logical to purchase "pre-owned" cars instead of buying new ones at full price. After proper research, one may also find a vehicle priced at an amount that is less than what it is worth, making it a very good deal. This paper discusses different machine learning models for predicting the selling price of a used car which can be used by both sellers and buyers to know at what best price a vehicle should be sold, after taking into consideration parameters such as the kilometers driven, in which year the vehicle was purchased, no. of owners, fuel type, the present price of the car and so on. This model hence is a novel approach to tackle a specific, although widely seen problem, as it allows the people that are interested in buying a preowned car to get the best deal for it by making an informed decision. In this paper, different Machine learning algorithms such as random forest, decision tree, support vector machine, and multi-linear regression are used for this analysis. Multi-linear regression works best among all specified models owing to the fact that is the most accurate of all the models studied and has the highest accuracy percentage of 84.48%. In order to visualize the model performance with respect to the actual value, the graph is plotted.