Car Price Prediction Using Ai

e-ISSN: 2582-5208
International Research Journal of Modernization in Engineering Technology and Science

( Peer-Reviewed, Open Access, Fully Refereed International Journal )
Volume:04/Issue:02/February-2022 Impact Factor- 6.752 www.irjmets.com
CAR PRICE PREDICTION USING MACHINE LEARNING TECHNIQUES

Abishek R*1
*1Student, Big Data Analytics, SRM University, Chennai, Tamil Nadu, India.
ABSTRACT
As a result of incredible technological advancements and research of new technical expertise and huge
economical growth of our country, people started to buy cars more than other vehicles. Therefore, there arises
an enormous demand for cars, as the demand increases for new cars the used car market also booms alike. But
the used car market is highly manipulated by few numbers of people who govern the rates of the used cars and
also online selling websites designate the values for the used cars. This paper tries to study and investigate the
trends in used car prices and predicts the price of used cars with the help of supervised machine learning
algorithms. And to suggest which machine learning algorithm performs well among the selected methods for
predicting the cars price. There has been related work done with machine learning algorithms like linear
regression, multiple regression, random forest and so on. We wanted to study which algorithm predicts the car
price more reliably and accurately. So that this solution will be helpful for first time used car buyers and also for
sellers for determining the selling cost of the car.
Keywords: Supervised Machine Learning Algorithms, Linear Regression, Multiple Regression, Random Forest.
I. INTRODUCTION
In India the automobile market is a biggest business for international and Indian automobile companies. As the
boom and demand for automobiles increase there is also a big market opening for used cars. The used car
market is being manipulated and controlled by some of the online advertisement websites like olx and quickr,
but customers who want to buy a used car is easily being manipulated and cheated to a higher price which the
car isn’t worth buying for. I would like to propose a solution for this problem by using the help of artificial
intelligence and machine learning by using some supervised learning machine learning techniques and
algorithms to predict the used car prices based on some parameters[2]. And I want to investigate and compare
the accuracy which different algorithms produce on testing and predicting with the used car data. During 2019-
20 the entire automobile production in India was 26,353,293, But in 2020-21 the automobile production in
India was 22,652,108. We can see that there is a huge decline in automobile industry, people are preferring
more on used and second-hand vehicles than new vehicles. Therefore the system of used cars must be
standardized and a clear pricing system needs to be implemented. This paper suggests few machine techniques
which can be used to predict the prices of used cars with historical used car prices data and considering a mean
value from the list of prices for a specific car and assigning it as the predicted price for the given features and
parameters[3]. There has been many related work done regarding this topic and field but only very few or one
or two authors have done for Indian dataset, Thus I wanted to find a solution for this problem and find
prediction method to give the prices for used cars in a correct method. The data for this car price prediction
experiment is taken from various sources like Kaggle, web-scrapping and open source data websites which
provide free data. A car price prediction has been a high-interest research area, as it requires noticeable effort
and knowledge of the field expert. Considerable number of distinct attributes are examined for the reliable and
accurate prediction. As the demand for cars increase the demand for second hand and used cars also increases
so due to this high demand, we need to build a AI solution for solving this demand in a customer friendly way.
The customers are getting cheated and tricked for a higher price for a less worth used car if the customer wants
to buy it from a dealer who sells used cars. The dealer tries to sell a damaged or repaired car for high price to
customers who don’t know much about buying cars and stuff. The customer who doesn’t know about the
technical specifications and other prices of spare parts and how to deduct the price will easily be cheated with
high price. I wanted to solve this type of problem where the customer has to know exact price the car is worth
for. Using machine learning it is possible to predict the correct and worthy price for a given used car based on
previous data from various sellers and buyers. This can be done by training the model using used cars dataset
which has several features and parameters such as year of manufacturing, model year, number of cylinders,
number of kms/miles driven, diesel or petrol, automatic or manual or other type of transmission, the gearing
www.irjmets.com @International Research Journal of Modernization in Engineering, Technology and Science

[45]
e-ISSN: 2582-5208
system of the cars, the number of owners of the car etc., like this there are many features from which the cars
price can be predicted. And also we can add if there is any damage or is it flood affected or accidental damaged
car these factors can also be considered for predicting the correct and exact price of the car.
II. METHODOLOGY
I have selected the required used car prices dataset with needed features and parameters from Kaggle. Kaggle
is an open source Machine learning and data science platform which offers data and notebooks for data
scientists and data analysts. The required data is cleaned and pre-processed used machine learning techniques
before applying any algorithm for predicting the price. Then after pre-processing and cleaning the data first we
need apply train test split to keep the data into two parts for training and validation using train and test data
respectively. Then we must apply a simple linear regression model and predict the output and test its test and
train accuracy with the help of roc auc score then again, we need to train and test it with multiple linear
regression model and validate its accuracies. Then we need to use clustering methods and logistic regression
methods and knn methods for predicting the output of car price. Also, we can use random forests and decision
tree algorithms. At last, we need to compare all the accuracies of all the machine learning algorithms and
choose the best algorithms for the prediction.
III. MODELING AND ANALYSIS
Between the following examples, we do some linear regression.
Figure 1: Comparisons of price and density

[46]
e-ISSN: 2582-5208
Figure 2: Box plot between various fuel types of cars.
Figure 3: Box plot between various years of cars.

IV. RESULTS AND DISCUSSION
From the pair plot we can understand that the most affecting features are price, engine, power, mileage, fuel
type, transmission types. So we need to build the model based on those features. This project mainly aims to
develop a solution for car price prediction for customers and also sellers who sell cars online and as well as
offline, The used car prices are predicted using kms, transmission, owner type, mileage, cc of the engine and
various features. These prices are accurate and are trust worthy for the customers, further this car be
developed with hyperparamter tuning and by using neural networks and other deep learning algorithms.
[47]
e-ISSN: 2582-5208
Figure 4: Pair plot between all the features.
Figure 5: Accuracy of linear regression model

[48]
e-ISSN: 2582-5208
Figure 6: Accuracy of random forest
Figure 7: Accuracy and errors of random forest model

When comparing the accuracies and error scores of linear regression model and random forest we can infer
than the random forest model performs far better with high accuracy score of 91%. Therefore it is better to use
random forest model for car price prediction project.
V. CONCLUSION
I have selected the required used car prices dataset with needed features and parameters from Kaggle. Kaggle
is an opensource Machine learning and data science platform which offers data and notebooks for data
scientists and data analysts. The required data is cleaned and pre-processed used machine learning techniques
before applying any algorithm for predicting the price. Then after pre-processing and cleaning the data first we
need apply train test split to keep the data into two parts for training and validation using train and test data
respectively. Then we must apply a simple linear regression model and predict the output and test its test and
train accuracy with the help of roc auc score then again, we need to train and test it with multiple linear
regression model and validate its accuracies. Then we need to use clustering methods and logistic regression
methods and knn methods for predicting the output of car price. Also, we can use random forests and decision
tree algorithms. At last, we need to compare all the accuracies of all the machine learning algorithms and
choose the best algorithms for the prediction.
VI. REFERENCES
[1] NATIONAL TRANSPORT AUTHORITY. 2014. Available from:
http://nta.gov.mu/English/Statistics/Pages/Archive.aspx [Accessed 15 January 2014]
[2] MOTORS MEGA. 2014. Available from: http://motors.mega.mu/news/2013/12/17/auto-market8-
decrease-sales-newcars/ [Accessed 17 January 2014].
[3] LISTIANI, M., 2009. Support Vector Regression Analysis for Price Prediction in a Car Leasing

[49]
e-ISSN: 2582-5208
Application. Thesis (MSc). Hamburg University of Technology.
[4] Oprea, C, Making the decision on buying second-hand car market using data mining techniques
(Special, 2010), pp.17-26.
[5] C Ozgur, Z Hughes, G Rogers and S Parveen, Multiple Linear Regression Applications Automobile
Pricing (International Journal of Mathematics and Statistics Invention, 2016), pp.01-10
[6] Lessmann, Stefan, M. Listiani, and S. Voß, Decision support in car leasing: A forecasting model for
residual value estimation (2010).
[7] G.Chandrashekar and F. Sahin, “A survey on feature selection methods,” Computers Electrical
Engineering, vol. 40, no. 1, pp. 16–28, 2014. [Online]. Available:
http://www.sciencedirect.com/science/article/pii/S0045790613003066
[8] M.C.Newman,“Regression analysis of log-transformed data: Statistical bias and its correction,”
Environmental Toxicology and Chemistry, vol. 12, no. 6, pp. 1129–1133, 1993. [Online]. Available:
http://dx.doi.org/10.1002/etc.5620120618
[9] R.Taylor, “Interpretation of the Correlation Coefficient: A Basic Review,” Journal of Diagnostic Medical
Sonography, vol. 6, no. 1, pp. 35–39, 1990
[10] Sameerchand Pudaruth, “Predicting the Price of Used Cars using Machine Learning Techniques”;(IJICT
2014)
[11] Enis gegic, Becir Isakovic, Dino Keco, Zerina Masetic, Jasmin Kevric, ”Car Price Prediction Using
Machine Learning”; (TEM Journal 2019)
[12] Ning sun, Hongxi Bai, Yuxia Geng, Huizhu Shi, “Price Evaluation Model In Second Hand Car System
Based On BP Neural Network Theory”; (Hohai University Changzhou, China)
[13] Nitis Monburinon, Suwat Rungpheung, Sabir Buya, Pitchayakit Boonpou, “Prediction of Prices for Used
Car by using Regression Models” (ICBIR 2018).

[50]

Car Price Prediction Using Ai

Uploaded by

Copyright:

Available Formats

Car Price Prediction Using Ai

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Car Price Prediction Using Ai

Uploaded by

Copyright:

Available Formats

e-ISSN: 2582-5208

International Research Journal of Modernization in Engineering Technology and Science

CAR PRICE PREDICTION USING MACHINE LEARNING TECHNIQUES

www.irjmets.com @International Research Journal of Modernization in Engineering, Technology and Science

Figure 1: Comparisons of price and density

www.irjmets.com @International Research Journal of Modernization in Engineering, Technology and Science

Figure 2: Box plot between various fuel types of cars.

Figure 3: Box plot between various years of cars.

Figure 4: Pair plot between all the features.

Figure 5: Accuracy of linear regression model

www.irjmets.com @International Research Journal of Modernization in Engineering, Technology and Science

Figure 6: Accuracy of random forest

Figure 7: Accuracy and errors of random forest model

www.irjmets.com @International Research Journal of Modernization in Engineering, Technology and Science

www.irjmets.com @International Research Journal of Modernization in Engineering, Technology and Science

You might also like