Abstract:
The process of finding out a used automobile's list price is challenging since there
are so many variables that influence the used car market pricing. The goal of this
work is to create a supervised machine learning model that can properly forecast a
used car's price based on its features so that consumers can make an educated
decision. In actuality, the vendor is equally ignorant about the automobile's
current value and the asking price at which the car should be sold. As is common
knowledge, Ethiopia is one of the nation’s having a sizable used automobile
market. Ethiopia's largest region, Oromia, is also where used cars are most
prevalent. Customers looking to purchase a used automobile frequently struggle
to do their own car shopping and estimate the price of a certain used car within
their budget. Currently, Oromia lacks an online website service that can assist
buyers purchasing secondhand cars. In this thesis, we investigate this issue and
create a predictive model that aids prospective purchasers in determining the price
of used cars that they are interested in purchasing. The Oromia Transport
Agency's Finfinne headquarters is where the dataset was gathered. We converted
53732 pieces of data originally in excel format from the database to csv format.
We then employed 12,895 data samples for this study experiment after cleaning
the data. Data analysis that is exploratory has been done. Different machine
learning techniques were applied, including XGBoost, KNN, random forest, and
linear regression. Random forest was chosen as the top model after examining the
effectiveness of each model used in this study experiment. The chosen model had
a 99% success rate in correctly predicting the price. The model was then locally
launched as a web application for future user accessibility. The User Interface
acquires input from user and displays the Price of a car according to user’s inputs.
In general, this technique for estimating used car pricing will assist both buyers
and sellers of cars when making price predictions.
