⚠️ This repo has moved. It's now maintained as part ofhotel-intelligence-suite, a consolidated collection of hotel operations projects, with full commit history preserved. This repo is archived and kept read-only for reference.
End-to-end hospitality data science project — 5 years of hotel operations data (2022–2026) for The Assyrian Grand Hotel Melbourne.
Built by Khoshaba Odeesho | Assyrian AI
→ View Live on Streamlit Cloud
This project combines 13 years of real hotel operations experience with modern data science. Every KPI, every insight, and every recommendation reflects decisions made daily at the hotel front desk — now automated and visualised at scale.
| Layer | Details |
|---|---|
| Database | Supabase PostgreSQL · 7 tables · 4 views · 185,439 rows |
| Data Period | Jan 2022 – Dec 2026 · 1,826 daily records |
| ML Models | Linear Regression ✅ Champion · XGBoost |
| AI Engine | Google Gemini 2.0 Flash · Daily briefings |
| Dashboard | Streamlit · 6 pages · Live public URL |
| BI Report | Power BI · 4 pages · 10 DAX measures |
| Pipeline | n8n · 10-node daily automation |
The Assyrian Grand Hotel — Melbourne CBD
- 120 rooms across 4 categories (Standard, Deluxe, Executive, Suite)
- 5 years of simulated operations data based on real Melbourne market conditions
- Melbourne events modelled: F1 Grand Prix, AFL Finals, Melbourne Cup, public holidays
| KPI | 2022 | 2026 | Change |
|---|---|---|---|
| Occupancy Rate | 62.6% | 77.2% | +14.6pp |
| ADR | $606 | $793 | +31% |
| RevPAR | $385 | $617 | +60% |
| GOPPAR | $178 | $285 | +60% |
| Annual Revenue | $17.8M | $28.2M | +58% |
| Guest Satisfaction | 7.84 | 8.48 | +0.64 |
| Event | Avg RevPAR | vs Weekday |
|---|---|---|
| Boxing Day | $1,088 | +145% |
| New Year's Day | $1,084 | +144% |
| AFL Grand Final | $1,078 | +143% |
| F1 Grand Prix | $1,059 | +139% |
| Melbourne Cup | $1,014 | +128% |
| Model | Occupancy MAE | RevPAR MAE | RevPAR R² |
|---|---|---|---|
| Linear Regression ✅ | 3.91% | $52.42 | 0.8124 |
| XGBoost | 5.07% | $82.30 | 0.5480 |
Linear Regression outperformed XGBoost — hotel seasonality follows strongly linear patterns (weekend premiums, events, year-on-year growth). This demonstrates model selection judgement over algorithm defaults.
hotel_rooms (120 rows) — Room inventory, types, rates hotel_guests (4,500 rows) — Guest profiles, segments, loyalty hotel_bookings (171,944 rows) — 5 years of all bookings hotel_financials (1,826 rows) — Daily revenue and cost breakdown hotel_reviews (5,223 rows) — Guest satisfaction across 6 platforms hotel_kpi_daily (1,826 rows) — Pre-calculated daily KPIs hotel_ai_insights — AI-generated reports (n8n pipeline) Views: hotel_monthly_kpis — Monthly aggregated KPIs hotel_events_impact — Melbourne events vs RevPAR hotel_channel_performance — OTA vs direct analysis hotel_guest_segment_performance — Revenue by guest type
| Layer | Tool |
|---|---|
| Database | Supabase PostgreSQL |
| Data Generation | Python · pandas · numpy |
| ML Models | scikit-learn · XGBoost |
| AI Narratives | Google Gemini 2.0 Flash |
| Dashboard | Streamlit · Plotly |
| BI Report | Power BI · DAX |
| Automation | n8n · 10-node pipeline |
| Version Control | Git · GitHub |
hotel-intelligence/ ├── hotel_dashboard.py — Streamlit 6-page dashboard ├── hotel_ml_forecast.py — Linear Regression + XGBoost training ├── hotel_generate_data.py — 5-year realistic data generation ├── requirements.txt — Python dependencies ├── .gitignore — Excludes .env and secrets └── README.md — This file
git clone https://github.com/Assyrian91/hotel-intelligence.git
cd hotel-intelligence
pip install -r requirements.txt
# Create .env with Supabase credentials
# HOTEL_DB_HOST=your-pooler-host
# HOTEL_DB_PORT=6543
# HOTEL_DB_NAME=postgres
# HOTEL_DB_USER=postgres.your-project-ref
# HOTEL_DB_PASSWORD=your-password
streamlit run hotel_dashboard.pyAll data is synthetically generated using real Melbourne hotel market parameters — seasonality indices, event premiums, post-COVID recovery curves, and channel mix ratios based on published REIV and STR Global benchmarks for Melbourne CBD 4-star properties.
Khoshaba Odeesho
Data Analyst | AI Automation Engineer | 13 Years Hotel Operations
Melbourne, Australia
Built as part of a professional data science portfolio — demonstrating end-to-end capability from raw data to deployed AI product, with domain expertise that no other candidate can replicate.