# 🌍 Global Affordability Dashboard ### Urban Cost of Living Analysis - Tableau & Machine Learning [](https://www.python.org/) [](https://public.tableau.com/app/profile/tejashwini.saravanan8751/viz/finalprojectassigmentTableau/GlobalAffordabilityTracker2025) [](https://scikit-learn.org/) [](https://github.com/TejashwiniSaravanan/global-affordability-dashboard-tableau-ml) [](LICENSE)
**In an era of record-high inflation and wage stagnation, understanding the *real* cost of living is critical.** This project integrates global cost-of-living metrics with salary data across **3,700+ cities** to compute a custom **Affordability Score** - revealing where people thrive and where they barely survive.
[](https://public.tableau.com/app/profile/tejashwini.saravanan8751/viz/finalprojectassigmentTableau/GlobalAffordabilityTracker2025)

---

⚡ Key Results at a Glance

Metric	Result
🏙️ Cities Analysed	3,700+
🌐 Countries Covered	195+
🤖 Random Forest R² Score	0.975
📉 Model RMSE	1.13
🔑 Top Affordability Driver	Median Salary (+9.43 coeff)
🚨 “Crisis Zone” Countries	127 (High Cost - Low Salary)
🏆 Most Affordable City Type	Small metros with rent-to-salary < 5%

---

📊 Interactive Dashboard Suite

Click any dashboard title to explore the live, interactive Tableau version.

🌍 Dashboard 1: Global Affordability Tracker 2025	📈 Dashboard 2: Country-Level Insights
Visualises geographic “Economic Red Zones” where living costs exceed local earnings. Uses a custom Affordability Score to identify global disparities.	Benchmarks average salaries against total essential costs (rent, groceries, utilities) to reveal which nations face the highest inflation pressure.
🏙️ Dashboard 3: City Deep-Dive	💡 Dashboard 4: Key Insights
A granular analysis of the Top 10 and Bottom 10 cities. Highlights the “Rent Trap” where housing costs consume more than 40% of median salary.	Synthesises regional cost patterns and visualises K-Means clustering results to categorise cities into distinct economic profiles.

---

🛠️ Tech Stack

Category	Tools
Language	Python 3.x
Data Wrangling	Pandas, NumPy
Machine Learning	Scikit-Learn - Linear Regression, Random Forest, K-Means Clustering
Visualisation	Tableau Desktop / Public, Matplotlib, Seaborn
Project Management	MS Project, Notion, Google Workspace
Notebook	Jupyter Notebook

---

⚙️ Analytical Pipeline

1. 🧹 Data Wrangling & Feature Engineering

• Integration: Merged two Kaggle datasets - Global Cost of Living and Global Salary Data - using a pd.merge() inner join on City and Country.
• Standardisation: Resolved naming inconsistencies (e.g., "USA" vs "United States") and handled missing salary values via proxy imputation.
• Feature Engineering: Derived a custom Affordability Score metric:

\[\text{Affordability Score} = \frac{\text{Monthly Median Salary}}{\text{Rent} + \text{Groceries} + \text{Utilities}}\]

Most cities cluster around median affordability - proving that true financial comfort is reserved for a global minority.

---

2. 🤖 Machine Learning Models

A. Predictive Modelling - Random Forest Regressor

• Test R²: 0.975 - capturing complex, non-linear relationships between salary and essential costs.
• RMSE: 1.13 - significantly outperforming a single Decision Tree baseline.

The Random Forest Regressor achieved a near-perfect fit on held-out test data.

---

B. Key Drivers - Linear Regression (Standardised Coefficients)

Driver	Coefficient	Direction
💰 Median Salary	+9.43	✅ Strongest positive driver
🛒 Grocery Prices	-4.29	❌ Primary drag on affordability
🏠 Rent	-4.11	❌ Second largest negative driver

Identifying Median Salary, Groceries, and Rent as the primary drivers of the Affordability Score.

---

C. Market Segmentation - K-Means Clustering (k=5)

Cities were segmented into 5 distinct economic profiles:

Cluster	Profile	Countries
Cluster 1	High Salary - High Cost (manageable rent burden)	19
Cluster 0	High Cost - Low Salary (“Crisis Zone”)	127
Cluster 3	Low Salary - Low Cost	104

Rent as a percentage of salary varies dramatically across city clusters - revealing stark global inequality.

---

📊 Global Insights & Quartile Analysis

• Q1 (Lowest Affordability): Residents spend approximately 50% of their salary on essentials - a survival-based economy with no financial cushion.
• The Rent Lever: Rent burden falls by nearly 28 percentage points when moving from the lowest (Q1) to the highest (Q4) affordability quartile.
• Top Performers: Non-major metro areas like Emporia (USA) and Treherbert (UK) achieved the highest scores due to exceptionally low rent-to-salary ratios (~3-5%).

---

🚀 Business & Policy Recommendations

1. 🏠 Housing Policy Implement rent control or housing subsidies in any city where the Rent-to-Salary ratio exceeds 35%.

2. 💼 Corporate Strategy Businesses in high-cost cities should adopt Cost-of-Living-Adjusted (COLA) salaries to ensure wage fairness and improve talent retention.

3. 🚚 Supply Chain Optimisation Focus on local supply chain development for essential goods in “High Cost - Low Salary” cities to drive down grocery price indices.

---

💻 Getting Started

#1. Clone the repository
git clone https://github.com/TejashwiniSaravanan/global-affordability-dashboard-tableau-ml.git
cd global-affordability-dashboard-tableau-ml

#2. Install dependencies
pip install -r requirements.txt
 
#3. Launch the notebook
jupyter notebook cost-of-living-crisis.ipynb

To explore the Tableau dashboards, open cost-of-living-crisis.twb in Tableau Desktop, or visit the live Tableau Public version.

---

📂 Repository Structure

global-affordability-dashboard-tableau-ml/
│
├── 📁 images/                          # Dashboard screenshots & analytical plots
├── 📄 Executive_Summary.pdf            # 1-page business impact summary
├── 📊 Project_Presentation.pdf         # Full visual slide deck
├── 📓 cost-of-living-crisis.ipynb      # Python notebook - data cleaning & ML models
├── 🗄️  cost-of-living-crisis.csv        # Cleaned & merged global dataset (3,700+ cities)
├── 📈 cost-of-living-crisis.twb        # Tableau Workbook file
├── 📋 requirements.txt                 # Python dependencies
├── 📖 README.md                        # Project documentation (you are here)
└── ⚖️  LICENSE                          # MIT License

---

📝 Project Reflection

✅ What Went Well	Successfully built a robust predictive framework (R² = 0.975) and transformed complex global data into four user-friendly, interactive Tableau dashboards.
⚠️ What Didn’t Go Well	Missing and inconsistent data for smaller cities required proxy imputation, which may affect localised precision.
🔭 Future Work	Integrate real-time datasets and additional cost categories - Healthcare, Transport, and Education - for a 360° affordability view.

---

👤 Author

**Tejashwini Saravanan** [](https://www.linkedin.com/in/tejashwinisaravanan/) [](https://github.com/TejashwiniSaravanan) [](https://tejashwinisaravanan.github.io/global-affordability-dashboard-tableau-ml/)

---

📄 License

This project is licensed under the MIT License - see the LICENSE file for details.

---

*If you found this project insightful, please consider giving it a ⭐ - it helps others discover it!*