lakshanga-dev/proppredict-ai

How lakshanga-dev/proppredict-ai works

PropPredict is a two-market real-estate estimation prototype: users enter India or US property details in the HTML dashboard, receive a model estimate and calibrated approximate 80% interval, and can use related scenario and bulk tools. The API validates requests and applies the shared feature pipeline before inference. Separate training modules build market-specific stacked models. The graph also preserves documented dashboard capabilities; unsampled frontend behavior and API internals are represented only where README or excerpts support them. Estimates trained on synthetic demo data are not reliable market valuations.

Main entry points: api.py Single prediction · feature_engineering.py India features · model_india.py India model training · model_usa.py US model training

Components 17

Dashboard

HTML dashboard index.html

API service

Feature pipeline

India features feature_engineering.py
US features feature_engineering.py
Category encoding feature_engineering.py
India data loading feature_engineering.py
US data loading feature_engineering.py

Market models

India model training model_india.py
US model training model_usa.py
India model bundle india_model.joblib
US model bundle us_model.joblib

Other components

Property user

Connections 24

Property user → HTML dashboard: enters property README.md
HTML dashboard → FastAPI service: requests estimates README.md
FastAPI service → Request validation: validates input api.py
FastAPI service → Single prediction: routes requests README.md
FastAPI service → CSV prediction: routes CSV README.md
FastAPI service → Scenario tools: routes scenarios README.md
Single prediction → India features: engineers India rows api.py
Single prediction → US features: engineers US rows api.py
CSV prediction → India features: engineers India rows api.py
CSV prediction → US features: engineers US rows api.py
India features → Category encoding: uses category maps api.py
US features → Category encoding: uses category maps api.py
Single prediction → India model bundle: predicts with api.py
Single prediction → US model bundle: predicts with api.py
CSV prediction → India model bundle: predicts with api.py
CSV prediction → US model bundle: predicts with api.py
Single prediction → SHAP explanations: requests explanations README.md
India data loading → India model training: supplies training data model_india.py
US data loading → US model training: supplies training data model_usa.py
India model training → India features: engineers training data model_india.py
US model training → US features: engineers training data model_usa.py
India model training → India model bundle: saves trained bundle
US model training → US model bundle: saves trained bundle
Scenario tools → HTML dashboard: returns results

Drawn by GitDiagram on November 10, 2026 from the repository's file tree, README and source files. Read as Markdown ·

How lakshanga-dev/proppredict-ai works

Main entry points: api.py Single prediction · feature_engineering.py India features · model_india.py India model training · model_usa.py US model training

Connections 24

Property user → HTML dashboard: enters property README.md
HTML dashboard → FastAPI service: requests estimates README.md
FastAPI service → Request validation: validates input api.py
FastAPI service → Single prediction: routes requests README.md
FastAPI service → CSV prediction: routes CSV README.md
FastAPI service → Scenario tools: routes scenarios README.md
Single prediction → India features: engineers India rows api.py
Single prediction → US features: engineers US rows api.py
CSV prediction → India features: engineers India rows api.py
CSV prediction → US features: engineers US rows api.py
India features → Category encoding: uses category maps api.py
US features → Category encoding: uses category maps api.py
Single prediction → India model bundle: predicts with api.py
Single prediction → US model bundle: predicts with api.py
CSV prediction → India model bundle: predicts with api.py
CSV prediction → US model bundle: predicts with api.py
Single prediction → SHAP explanations: requests explanations README.md
India data loading → India model training: supplies training data model_india.py
US data loading → US model training: supplies training data model_usa.py
India model training → India features: engineers training data model_india.py
US model training → US features: engineers training data model_usa.py
India model training → India model bundle: saves trained bundle
US model training → US model bundle: saves trained bundle
Scenario tools → HTML dashboard: returns results

Drawn by GitDiagram on November 10, 2026 from the repository's file tree, README and source files. Read as Markdown ·