Analysis of the UN's World Happiness Index with machine learning

Maaike de Jong
June 2020

Introduction

The World Happiness Report is an annual publication of the United Nations Sustainable Development Solutions Network. It presents the results of a survey of the state of global happiness that ranks 156 countries by how happy their citizens perceive themselves to be. The rankings of national happiness are based on a Cantril ladder survey, where nationally representative samples of respondents are asked to think of a ladder, with the best possible life for them being a 10, and the worst possible life being a 0. They are then asked to rate their own current lives on that 0 to 10 scale.

There are large differences in happiness among countries and continents, with usually the top happiest countries from Western Europe, and the least happy countries from sub-Saharan Africa and the Middle East. In this analysis I use machine learning to investigate which factors are important in determining the overall happiness score of nations. Specifically, I aim to answer the following questions:

1. Which economical, social and environmental factors contribute most to a country’s happiness?
2. How well can we predict happiness score with these factors?

Analysis

I use scikit-learn's linear regression model with recursive feature elimination to analyse the contribution of 20 World Development Indicators (WDIs) with the World Hapiness Score. I used the average Happiness score of the past five years, and started with the following WDIs per country for the past 10 years:

• Access to electricity (% of population)
• CO2 emissions (metric tons per capita)
• Compulsory education, duration (years)
• GDP growth (annual %)
• GDP per capita (current US$)
• Individuals using the Internet (% of population)
• Land area (sq. km)
• Life expectancy at birth, total (years)
• PM2.5 air pollution, mean annual exposure (micrograms per cubic meter)
• People using at least basic drinking water services (% of population)
• Population density (people per sq. km of land area)
• Population growth (annual %)
• Proportion of seats held by women in national parliaments (%)
• Refugee population by country or territory of origin
• Population, total
• Renewable energy consumption (% of total final energy consumption)
• School enrollment, primary (gross), gender parity index (GPI)
• Terrestrial protected areas (% of total land area)
• Urban population (% of total population)

Data

The World Happiness Report data from 2015-2019 can be found on Kaggle (five separate csv files).
I retrieved a csv file with 1400+ World Development Indicators for all countries from the World Bank's website (under 'Bulk Downloads').

File structure

The analysis is divided over four notebooks:

1. Data wrangling to combine the different datasets
2. Data cleaning and visual exploration
3. Regression analysis with scikit-learn
4. Visualising the data in graphs and world maps