Skip to content

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

6 Commits
 
 
 
 
 
 

Repository files navigation

🤖 AutoML for 🌍 World Happiness Prediction

📄 Abstract

This study 🧠 predicts national 😊 happiness scores by evaluating key 💰 socio-economic indicators, including 🏭 economic output, 👨‍👩‍👧‍👦 social support, 🏥 life expectancy, 🗽 individual freedoms, ⚖️ governmental integrity, and 🤲 societal generosity. Data from the World Happiness Report, published by the 🏢 United Nations Sustainable Development Solutions Network, enables predictive modeling through the ⚙️ H2O AutoML platform, revealing complex variable interactions.

The research explores several key 🔍 questions:

  1. 📊 How do 💰 socio-economic predictors explain national 😊 happiness?
  2. 🧩 Are core statistical assumptions (e.g., multicollinearity) violated?
  3. 🔗 How do predictors interact in complex multivariate contexts?
  4. ⭐ Which variables are the most significant contributors to 😊 happiness?
  5. 📈 Does regularization enhance model performance?
  6. ⚙️ What hyperparameters optimize predictive accuracy?
  7. 🌐 How generalizable is the model across diverse geopolitical contexts?

The study begins with 📚 dataset evaluation, followed by 🤖 machine learning model development using ⚙️ H2O AutoML. Predictive performance is assessed using advanced 🔬 statistical metrics.

🗂️ Dataset Information

🎯 Target Variable (Dependent Variable):

  • 😊 Happiness Score: Self-reported life satisfaction on a 0-🔟 scale, with 🔟 indicating maximum well-being.

📊 Predictor Variables (Independent Variables):

  1. 🌍 Country: Nation being evaluated.
  2. 📍 Region: Geopolitical classification.
  3. 🏆 Happiness Rank: Relative ranking by happiness score.
  4. 📐 Standard Error: Measurement variability.
  5. 💰 Economy (GDP per Capita): National economic output.
  6. 👨‍👩‍👧‍👦 Family: Social and familial support.
  7. 🏥 Health (Life Expectancy): Public health indicators.
  8. 🗽 Freedom: Perceived personal autonomy.
  9. 🚫 Trust (Government Corruption): Institutional integrity and absence of corruption.
  10. 🤲 Generosity: Charitable behavior and social goodwill.
  11. 🏚️ Dystopia Residual: Minimum theoretical happiness benchmark.

📏 Model Performance Metrics

Model performance is evaluated using several key 📈 statistical measures:

  • 📉 Mean Squared Error (MSE): Average prediction error.
  • 📊 Root Mean Squared Error (RMSE): Standard deviation of prediction errors.
  • 📐 Mean Absolute Error (MAE): Average prediction magnitude.
  • 🔢 Root Mean Squared Log Error (RMSLE): Log-scaled prediction error.
  • 📈 R-squared (R²): Explained variance proportion.

🌟 Feature Importance Analysis

The following features significantly impact 😊 happiness prediction:

Screenshot 2023-10-25 at 10 05 12 AM
  • 🏚️ Dystopia Residual: Most critical determinant.
  • 💰 Economy (GDP per Capita): Key economic indicator.
  • 🏥 Health (Life Expectancy): Vital well-being metric.
  • 🤲 Generosity: Reflects societal altruism.
  • 🚫 Trust (Government Corruption): Indicates public trust.

🔍 Model Assumptions & Insights

  • 📊 Predictive Significance: All variables meaningfully impact 😊 happiness.
  • 📐 Assumption Validity: Statistical tests confirm model assumptions.
  • 🧩 Multicollinearity: Managed through dimensionality reduction and regularization.
  • 🛠️ Regularization Impact: Enhances stability while reducing overfitting.

🤖 Model Training with ⚙️ H2O AutoML

The project used ⚙️ H2O AutoML for 🤖 model training, selecting a Generalized Linear Model (GLM) as the best-performing algorithm.

Screenshot 2023-10-25 at 10 08 50 AM

📊 Training Data Metrics:

  • 📉 MSE: 0.0006774
  • 📊 RMSE: 0.0260
  • 📐 MAE: 0.01989
  • 🔢 RMSLE: 0.0049
  • 📈 R²: 0.9995

📈 Cross-Validation Metrics:

  • Slightly lower scores on unseen data (📈 R² = 0.99), indicating strong generalization.

💻 Coding & Documentation Standards

The project follows best practices in 📊 data science, 💾 software engineering, and 🎓 academic research. The codebase is modular, scalable, and well-documented to ensure 📑 reproducibility and 🔬 methodological transparency.

🏁 Conclusion

The ⚙️ H2O AutoML framework demonstrates exceptional performance in predicting 😊 happiness scores using 💰 socio-economic indicators. The findings reveal complex relationships among economic, social, and political factors, offering significant insights for evidence-based 🌍 policy development aimed at improving global well-being.

📚 References

About

AutoML, short for Automated Machine Learning, is a set of techniques and tools that automate various tasks in the machine learning process.

Resources

Stars

0 stars

Watchers

1 watching

Forks

Releases

Packages

Contributors

Languages