Vietnamese House-Price Estimator

Assignment 01 — Intelligent System Development. Input → representation → model → prediction.
Model: HGB tuned (lr=0.1, it=800) · trained on 23,723 listings · test MAE 1.02 billion VND (R² 0.618).

Floor area (m²) — required, training range 15–400

Frontage (m) — optional, training range 1–75

Access road width (m) — optional, training range 1–85

Floors — optional, training range 1–8

Bedrooms — optional, training range 1–8

Bathrooms — optional, training range 1–9

Province

District

House direction

Balcony direction

Legal status

Furniture state

Blank fields are not stated, which the system encodes explicitly with a *_missing indicator rather than silently guessing. Estimates are for asking prices of mass-market properties in the range 1.0–11.5 billion VND; the training data is censored above that ceiling, so the system under-predicts luxury properties. This is coursework, not a valuation service.