Overview01
First-time student buyers shopping the $5 to 10k used-car market rarely have the automotive know-how to judge which cheap car is a safe bet. The goal of AutoRisk is to turn public safety data into a plain reliability score so they can buy with data instead of a gut feeling.
Plan and progress02
- Fetch (built). Pulls complaints, recalls, and investigation records per vehicle from the NHTSA API into structured CSVs, for 30 popular 2012 to 2014 models.
- NLP (next). Group free-text complaints into recurring failure patterns.
- Model (next). Estimate depreciation and value trajectory.
- Score and app (next). Combine complaint patterns, recall frequency, and investigations into one risk score behind a simple app.
Status03
In progress. The NHTSA ingestion pipeline is built. Complaint clustering, the depreciation model, scoring, and the app are next.
Tech
Pythonrequestspandas