PropreModelia: Land Price Predictor

PrototypeFull-Stack ML Web Application / 2024

Problem

Land valuation in Algeria lacks transparent price estimates based on location and land size.

Solution

Trained single-feature and multi-feature regressions on local land data, exposed them through FastAPI endpoints, and built a Next.js UI for DZD price estimates.

Key Features

Single-feature regression: predict price from superficie alone

Multi-feature regression: predict price from superficie + secteur (campagne / ville)

Dynamic form toggle between the two prediction modes

FastAPI backend with /predict-single and /predict-multi endpoints

Price output formatted in Algerian Dinar (DZD)

Responsive UI built with Material-UI and Tailwind CSS

Deployed on Vercel with a live preview

Architecture Map

Mermaid

flowchart LR
  User["Next.js Frontend
(Material-UI + Tailwind CSS)"] --> Single["Single-Feature Form
(superficie)"]
  User --> Multi["Multi-Feature Form
(superficie + secteur)"]
  Single --> S_API["POST /predict-single"]
  Multi --> M_API["POST /predict-multi"]
  S_API --> ML["scikit-learn
Linear Regression
(Python / FastAPI)"]
  M_API --> ML
  ML -.->|"predicted_price (DZD)"| User

PropreModelia Screenshots

3 views

PropreModelia app home screen with land price prediction form — The PropreModelia home screen: users enter land details and get an instant price estimate in Algerian Dinar.

Single-feature linear regression scatter plot: Superficie vs Prix with fitted prediction line — Single-feature model: linear regression on superficie (m2) vs price (DZD). The fitted line shows the learned price trend across the dataset.

Multi-feature regression scatter plot: Actual vs Predicted Prices by Secteur (ville and campagne) — Multi-feature model: regression by secteur. Secteur 1 (ville, blue/orange) commands higher prices than Secteur 0 (campagne, green/red) across all superficie values.