Description
2. The Data
The data to be considered for this project was downloaded from the World Happiness report of
2017; see
Statistical Appendix for ”The social foundation of world happiness”, John F. Helliwell, Hailing
Huang and Shun Wang, Chapter 2, World Happiness Report 2017 , March 21, 2017
The file happiness2017.csv in Quercus contains the data. The main variables in the dataset
are:
• Ladder- the happiness score or subjective well-being of a country. Values are in the range
0 to 10, where 10 represent the best possible life.
• LogGDP- the logarithm of a country’s 2016 GDP.
• Social- the national average of the binary responses to the question “If you were in trouble,
do you have relatives or friends you can count on to help you whenever you need them, or
not?”
• HLE- Healthy life expectancy at birth based on data reported from the World Health Organization (WHO)
1
• Freedom – the national average to the question “Are you satisfied or dissatisfied with your
freedom to choose what you do with your life?”
• Generosity – the residual of regressing the national average of the response to the question
“Have you donated money to charity in the past month?” on GDP per capita
• Corruption – the national average to the two questions “Is corruption widespread throughout the government or not?” and “Is corruption widespread within businesses or not?”
• Positive – a measure of positive affect as the average of three positive affect measures –
happiness, laugh and enjoyment in the Gallup World Poll.
• Negative – a measure of negative affect as the average of three negative affect measures –
worry, sadness and anger in the Gallup World Poll.
• gini- the gini of household income in international dollars as reported in Gallup
The aim is to predict the happiness score of a country using the other nine (9) variables. Your
report should include the following four sections, along with an Appendix.
3. The Project Report
I. Summary/Abstract (5 marks)- Write an executive summary of your analysis by using between 100 and 300 words. Do not include any plots in this part.
II. Data Manipulation and Summary (10 marks)- Set the seed of your randomization to be
the last four digits of your student number or your partner’s student number. From the raw
data, which is in alphabetical order, take a random sample of 100 countries for your data.
Carry out the natural first step to data analysis by displaying multivariate data graphically
and to obtain summary statistics. Note that, you should take into account any interesting
features in the data that you come across. This implies making suitable transformations,
handling outliers, removing countries with missing values or making suitable imputations for
missing values. As a next step, you should check multivariate normality assumption for your
data as most of the techniques we cover require normality.
III. Multiple Linear Model using Original variables (10 marks)- Fit a linear model for the
response variable- happiness score using all the original explanatory variables. Check whether
the assumptions of this model are satisfied. Report the ‘Adjusted R-squared’ value from the
R summary(lm(…)) output
IV. Multiple Linear Model using Principal Components (10 marks) – Fit a linear model
for the response variable- happiness score using less than 9 principal components. Discuss
whether you found the principal components from the covariance matrix or the correlation
matrix. Discuss how you decided on the number of principal components that you retained.
Produce a scree plot. Report the ‘Adjusted R-squared’ value from the R summary(lm(…))
output for your ‘final’ model. Interpret the principal components retained.
2
