Motorcycle Classifier
Machine Learning Pipeline
About the Project
This repository contains a complete machine learning pipeline designed to classify the worthiness of used motorbikes listed on the OLX Indonesia marketplace. The classification determines whether a used motorbike is worth buying based on a combination of price ratio, documentation completeness, motorbike age, mileage, and physical/engine condition keywords extracted from the listing descriptions.
Features & Details
Web Scraping Automation
Extracted thousands of motorbike listings dynamically from OLX Indonesia using headless Selenium.
Data Cleaning & Imputation
Replaced missing mileage values using statistical medians and cleaned inconsistent document strings.
Custom Heuristic Labeling
Designed a scoring algorithm based on price ratios, documents, conditions, and age to classify listings.
Classification Modeling
Trained multiple models to predict worthiness, achieving a peak 90.4% accuracy with Decision Trees.
Project Gallery
Missing Values
Target Distribution
Correlation Heatmap
Decision Tree CM
ROC Comparison
Price Distribution
Technical Details
Language
Python 3
Machine Learning
Scikit-Learn (Decision Tree, Random Forest)
Data Processing
Pandas, NumPy
Web Scraping
Selenium WebDriver
Visualization
Matplotlib, Seaborn