AI-Assisted Location Decision Support System (ALSDS)

Instructor Baseline Version

Live App

This repository provides the baseline application and infrastructure for the ALSDS capstone project.

It includes:

• Core data files for Worcester, MA (CBGs, POIs, visits, parameters)
• A baseline Huff-style model implementation (huff_engine.py)
• A Flask-based web application with:
  • Guided chatbot (Azure OpenAI / GPT-4o)
  • Interactive map (Leaflet + GeoJSON)
  • Model execution API
• Azure deployment configuration (App Service compatible)

Important Setup Instructions

This is NOT a fork-based workflow

Each team must:

• Create a NEW repository in your GitHub Organization
• Copy (clone or download/upload) this repository into your repo

Do NOT fork this repository.

Required Repository Naming

Each team repository must be named:

alsds-teamX-app

Example:

alsds-team3-app

Baseline Version

Stable infrastructure release: v1-baseline

All teams must start from this version before making any changes.

Do NOT Modify (Critical Infrastructure)

The following files must remain unchanged:

startup.sh requirements.txt (do not remove any current requirements, you can add more items if required.) app.py (initially, unless explicitly instructed) deployment configuration (GitHub Actions / Azure settings) environment variable naming API route structure (/api/run_huff, /api/ask)

These are required for:

• Azure deployment
• database connectivity
• OpenAI integration
• instructor testing

What Teams ARE Expected to Modify/Create

Teams should focus on:

• huff_engine.py
• database design (Azure SQL)
• data preprocessing / optimization
• optional UI enhancements

Huff Model Interface Requirement

Your implementation must preserve this function:

def run_huff_model(candidate_lat, candidate_lon, business_category, floor_area, db_connection):

Return format must include:

predicted_visits market_share competitors runtime_ms notes

You are free to completely redesign the internal logic.

Map Integration

The app uses:

static/data/worcester_cbgs_map.geojson

Ensure this file exists and is not removed.

The map supports:

• click-to-select candidate location
• competitor visualization

Chatbot Behavior

The chatbot will:

1. Guide the user to input:

• business category
• location (map click or coordinates)
• floor area

2. Run the model
3. Provide explanation
4. Answer follow-up questions

Deployment Workflow

After your repository is ready:

1. Instructor connects your repo to Azure Web App
2. GitHub Actions handles deployment
3. App is available at your Azure URL

Required Endpoints

Your deployed app must support:

/health → returns {"status":"ok"}

/dbcheck → verifies database connection

/ → loads UI (chat + map)

Evaluation Focus

Your project will be evaluated on:

• Model quality (Huff implementation)
• Database design and efficiency
• Query performance / runtime
• Code organization
• Ability to explain results (chatbot)

Notes

• Keep your repository clean and well-structured
• Commit changes regularly
• Create a dev branch and test locally before deployment