The goal of this project is to conduct an exploratory study in which prices of classic cars are estimated using machine learning methods.
The data for this analysis was scraped from the classic trader website, which describes itself as the international marketplace for classic vehicles. After cleaning and preprocessing, about 8 thousand observations are available for the analysis.
To estimate the car prices, following models were trained:
- Linear Regression
- Random Forrest
- K-nearest Neighbors
- Ensemble model(s) (averaging)
Unfortunately, the model cannot provide reliable prediction of the car prices. Probable causes are data quantity, data quality, interaction effects and omitted variable bias.
This project was created as part of the last course of the Professional Certificate in Data Science by HarvardX via edx.org.