Demographic-Service-Correlation

Overview

This project investigates the relationship between demographic factors and access to basic services in Indonesian provinces using Canonical Correlation Analysis (CCA). The analysis provides insights into the multidimensional relationships between population growth, density, and service accessibility (e.g., water, sanitation, and healthcare).

The study revealed:

• A strong canonical correlation (0.728 for the first function) between demographic variables and service accessibility.
• Key variables influencing these relationships include population density (demographic factor) and access to sanitation (service accessibility).

Files

• Code.Rmd: Contains the R code used for analysis, including data preprocessing, transformation, assumption testing, and CCA.
• Dataset_Kesehatan.xlsx: Dataset containing variables for population dynamics and service accessibility across Indonesian provinces.
• The Paper.docx: Detailed analysis report explaining methods, results, and interpretations.

Methodology

1. Data Preprocessing:

   • Variables were grouped into two categories:
     • Demographic Factors: Population growth rate, percentage, and density.
     • Access to Basic Services: Drinking water, sanitation, and healthcare.
   • Data transformations (log, square root, Box-Cox, Yeo-Johnson) were tested to ensure assumptions were met.

2. Assumption Testing:

   • Multivariate Normality:
     • Pre-transformation p-values indicated violation of normality (e.g., p < 0.05).
     • Post Yeo-Johnson Transformation: p-values for both groups satisfied normality assumptions.
   • Multicollinearity:
     • Variance Inflation Factor (VIF) values for all variables were below 10, indicating no multicollinearity.
   • Linearity:
     • Correlation analysis confirmed linear relationships post-transformation.

3. Canonical Correlation Analysis:

   • Canonical coefficients, weights, and loadings were calculated for each group.
   • Statistical tests:
     • Wilks’ Lambda (F-approximation) showed the first canonical correlation (p = 0.000877) was significant.

Key Results

Canonical Correlation Coefficients

• First Function: 0.728 (significant, p = 0.000877)
• Second Function: 0.427 (not significant, p > 0.05)
• Third Function: 0.063 (not significant, p > 0.05)

Variable Importance

• Demographic Factors:
  • Population Density: Most significant variable with a canonical weight of -0.179.
  • Population Growth Rate: Contributed less significantly (-0.065).
• Access to Basic Services:
  • Sanitation Access: Strongest contributor with a canonical weight of -0.110.
  • Healthcare Access: Moderate contribution (-0.080).

Loadings and Relationships

• Population density exhibited a strong negative correlation (-0.807 loading), indicating an inverse relationship with access to services.
• Sanitation access had the highest positive loading (0.909), emphasizing its importance in determining service accessibility.

Statistical Tests

• Wilks’ Lambda for the first canonical function was 0.383 with an F-approximation p-value of 0.000877.
• Remaining functions were not statistically significant (p > 0.05).

Interpretation

• High population density is strongly correlated with lower access to basic services, highlighting regional disparities.
• Improving sanitation and healthcare access is critical for mitigating the effects of demographic pressures in densely populated areas.

Usage

Requirements

Install the required R libraries using:

install.packages(c('expm', 'car', 'MASS', 'bestNormalize', 'MVN', 'mvShapiroTest', 'CCP', 'CCA', 'GGally'))

Running the Code

1. Open the Code.Rmd file in RStudio.
2. Ensure Dataset_Kesehatan.xlsx is in the correct path as specified in the code.
3. Run the code sequentially for preprocessing, assumption testing, and analysis.
4. Outputs include canonical coefficients, weights, and diagnostic plots.

Outputs

• Canonical coefficients for three functions.
• Weights and loadings for interpreting variable contributions.
• Assumption validation plots and correlation matrices.

Key Insights

• Canonical correlation analysis is a powerful method for exploring relationships between demographic factors and service accessibility.
• Policymakers can use these findings to prioritize resource allocation in regions with high population density and poor sanitation access.