About
A small end-to-end data product: scrape listings, enrich locations, train a pricing model, surface undervalued picks, with optional images for future vision models.
Daily automation (Windows)
Use Task Scheduler to run once per day in this order:
- 1
Scrape listings
python scripts/scrape_dotproperty.py - 2
Geocode & enrich
python scripts/geocode_enrich.py - 3
Train undervalued model
python scripts/train_undervalued_model.py
Run from the development folder. More scrape days improve 3m/6m price stats on the Statistics page.
Portfolio angle
Emphasize the pipeline (ethical scraping, rate limits, data quality, reproducible ML, clear UI) and what you'd add next: multi-category scrapers (condo / lot), proper evaluation, and calibrated vision labels.