dataaanlysisfinalproject

[dataaanlysisfinalproject]

Predicting GDP Per Capita Using Development Indicators

Results

Descriptive Statistics:

GDP per capita: Mean = $14,751, Std Dev = $21,612, Min = $244.2, Max = $149,161

Fixed broadband subscriptions: Mean = 11.6, Std Dev = 12.2, Min = 0.00016, Max = 43.2

Improved water source: Mean = 88.3%, Std Dev = 14.5%, Min = 39.9%, Max = 100%

Internet users: Mean = 39.9, Std Dev = 28.0, Min = 0.00, Max = 96.2

Mortality rate (under-5): Mean = 36.0, Std Dev = 35.5, Min = 2.1, Max = 172

Women in parliament: Mean = 19.2%, Std Dev = 10.7%, Min = 0.00%, Max = 56.3%

Rural population: Mean = 42.2%, Std Dev = 23.5%, Min = 0.00%, Max = 91.2%

Urban population: Mean = 57.8%, Std Dev = 23.5%, Min = 8.8%, Max = 100%

Birth rate: Mean = 21.4, Std Dev = 10.4, Min = 8.2, Max = 49.9

Bivariate Analyses:

Strongest Correlations:

Fixed broadband subscriptions: r = 0.789, p < .0001

Internet users: r = 0.776, p < .0001

Log of under-5 mortality rate: r = -0.710, p < .0001

Urban population rate: r = 0.619, p < .0001

Weakest Correlations:

Proportion of women in parliament: r = 0.253, p < .001

Rural population rate: r = -0.619, p < .0001

Multivariable Analyses:

Lasso Regression Results:

Retained Predictors: Fixed broadband subscriptions, internet users, under-5 mortality rate, urban population rate, birth rate, improved water source

Regression Coefficients:

Fixed broadband subscriptions: 9587

Internet users: 9030

Under-5 mortality rate: 397.5

Urban population rate: 2329

Birth rate: 4875

Improved water source: -31.82

Model Performance:

Training data R-square: 0.6827

Test data R-square: 0.6806

Mean squared error (training): 124,390,237

Mean squared error (testing): 123,487,565

[dataaanlysisfinalproject]

Predicting GDP Per Capita Using Development Indicators

Methods

Sample: The sample consists of data from the World Bank, covering 248 countries for the years 2012 and 2013. Only data from 2012 was used for the primary analysis, with 2013 data reserved for validation. The dataset includes national, regional, and global estimates, with valid observations for a minimum of 190 countries per variable.

Measures:

Response Variable: GDP per capita (current US$)

Predictors:

Fixed broadband subscriptions (per 100 people)

Access to improved water sources (% of population)

Internet users (per 100 people)

Mortality rate for under-5-year-olds (per 1,000)

Proportion of seats held by women in national parliaments (%)

Rural population rate (% of total population)

Urban population rate (% of total population)

Birth rate (crude, per 1,000 people)

Statistical Analyses:

Descriptive Statistics: Calculated mean, standard deviation, minimum, and maximum for all variables.

Bivariate Analyses: Used Pearson correlation to test associations between predictors and GDP per capita. Log transformations were applied where necessary to linearize relationships.

Multivariable Analyses: Employed lasso regression with least angle regression selection algorithm to identify the best subset of predictors. Cross-validation with 10 folds was used to determine the optimal model. Predictive accuracy was assessed using mean squared error rates on training and test datasets.

[dataaanlysisfinalproject]

Predicting GDP Per Capita Using Development Indicators

Research Question: What are the best predictors of a country's GDP per capita among various development indicators?

Motivation/Rationale: Having spent my career in developing countries working for an international aid agency, I have witnessed first hand how small income boosts can dramatically improve the quality of life in impoverished communities. By identifying which development goals most effectively lead to increases in GDP per capita, aid agencies globally can better focus their activities towards improving those indicators, indirectly leading to increased incomes for families.

Implications: Understanding the key predictors of GDP per capita can help aid agencies allocate resources more effectively, ensuring that development efforts have the greatest impact on improving living standards. This research aims to provide empirical evidence to guide these decisions, ultimately contributing to more effective and targeted development strategies.