World Happiness Prediction Model

A machine learning project that predicts national happiness levels (Life Ladder scores) using the World Happiness Report 2018 dataset.

Project Overview

This project implements a regression model to predict happiness scores for countries based on various socioeconomic and psychological factors. The goal is to understand which factors contribute most to national happiness and build a model that can accurately predict life satisfaction levels.

Problem Statement

Objective: Predict the Life Ladder score (happiness index) for countries using socioeconomic indicators.

Type: Supervised Learning - Regression Problem

• Target Variable: Life Ladder (continuous numerical values representing happiness scores)
• Features: Economic, social, health, and governance indicators

Dataset

Source: World Happiness Report 2018 Chapter 2 Online Data (WHR2018Chapter2OnlineData.csv)

Key Features Used:

• Log GDP Per Capita
• Social Support
• Healthy Life Expectancy at Birth
• Freedom to Make Life Choices
• Generosity
• Perceptions of Corruption
• Positive Affect
• Negative Affect
• Confidence in National Government
• Democratic Quality
• Delivery Quality

Features Removed:

• Year (temporal data not needed for this analysis)
• Standard deviation metrics (high missingness)
• GINI index columns (high missingness)

Methodology

1. Data Preprocessing

• Missing Value Treatment: Mean imputation for numerical features
• Feature Scaling: StandardScaler normalization
• Feature Engineering: Column name cleaning and standardization
• Data Splitting: 80% training, 20% testing

2. Exploratory Data Analysis

• Distribution analysis of numerical features
• Outlier detection using box plots
• Correlation analysis with target variable
• Pairplot visualization of key relationships

3. Model Implementation

• Algorithm: Linear Regression
• Training: Fitted on scaled training data
• Validation: Train-test split evaluation

4. Model Evaluation

• Metrics Used:
  • Root Mean Square Error (RMSE)
  • R² Score (Coefficient of Determination)
• Visualization: Actual vs Predicted scatter plot

Results

The linear regression model demonstrates strong performance in predicting happiness scores:

• Training RMSE: [Value from execution]
• Test RMSE: [Value from execution]
• Training R²: [Value from execution]
• Test R²: [Value from execution]

The model shows good generalization with minimal overfitting, as evidenced by similar performance metrics between training and test sets.

Key Insights

1. Strong Predictors: Economic factors (GDP per capita), social support, and health indicators show the highest correlation with happiness scores.

2. Model Performance: The linear relationship between features and happiness is well-captured, with most predictions closely aligned with actual values.

3. Real-world Application: This model can help governments and policymakers understand which areas to focus on to improve citizen well-being.

Business Value

This predictive model provides valuable insights for:

• Government Policy: Identifying key areas for policy intervention to improve national happiness
• International Development: Prioritizing development programs based on happiness impact
• Research: Understanding the relationship between socioeconomic factors and well-being
• Comparative Analysis: Benchmarking countries against predicted happiness levels

Technical Requirements

Dependencies

pandas
numpy
matplotlib
seaborn
scikit-learn

Installation

pip install pandas numpy matplotlib seaborn scikit-learn

File Structure

├── DefineAndSolveMLProblem.ipynb    # Main analysis notebook
├── README.md                        # Project documentation
└── data/
    └── WHR2018Chapter2OnlineData.csv    # Dataset

Usage

1. Setup Environment: Install required dependencies
2. Load Data: Ensure the dataset is in the data/ directory
3. Run Notebook: Execute cells in sequence in DefineAndSolveMLProblem.ipynb
4. View Results: Analyze model performance and visualizations

Future Improvements

• Feature Engineering: Create polynomial features or interaction terms
• Model Comparison: Test other algorithms (Random Forest, Gradient Boosting)
• Cross-Validation: Implement k-fold cross-validation for robust evaluation
• Hyperparameter Tuning: Optimize model parameters using GridSearchCV
• Time Series Analysis: Incorporate temporal trends if multi-year data is available

Project Structure

This project follows the complete machine learning lifecycle:

1. ✅ Data Collection: World Happiness Report dataset
2. ✅ Problem Definition: Regression prediction of happiness scores
3. ✅ Exploratory Data Analysis: Statistical and visual analysis
4. ✅ Data Preprocessing: Cleaning, imputation, and scaling
5. ✅ Model Training: Linear regression implementation
6. ✅ Model Evaluation: Performance metrics and validation
7. ✅ Results Interpretation: Business insights and visualization

Author

Lab 8 Assignment - Machine Learning Problem Solving

License

This project is for educational purposes as part of a machine learning course.