The main goal was to create a complete dataset to calculate accurate global rankings. Many countries do not report all 30 indicators, which creates gaps in the raw data. Missing data makes it impossible to compare nations fairly. The objective was to fill these gaps without changing the true shape of the data. The input data came from various global reports containing missing values across different categories. I needed to test different mathematical methods to guess the missing numbers. The final aim was to choose a method that kept the original data patterns intact and then apply a standard scale to all indicators to calculate the final scores.

The project successfully produced a complete and standardized dataset for all 126 countries. Using the MissForest model ensured the filled data closely matched real world patterns without creating extreme errors. The normalization process allowed for a direct comparison across completely different indicators, such as comparing tax rates to health scores. The final outcome is a clear ranking system divided into specific performance categories. This organized structure allows government leaders to easily see their strengths and areas needing improvement. The clean dataset now serves as a reliable foundation for future policy analysis and international comparisons.

I tested three methods to fill the data gaps. Multiple Imputation by Chained Equations borrows values from a similar observed country to keep the numbers realistic. Weighted K-Nearest Neighbors groups regional peers, assuming close countries share similar traits. MissForest uses an Iterative Random Forest model to predict missing values. This third model maps out difficult relationships, such as how national wealth changes health scores across different income levels. I chose the MissForest method because the stakeholders needed the final numbers to look very similar to the original data shape. After fixing the missing data, I used a standard statistical process called normalization. This step converted every indicator to a simple scale from zero to one hundred. This standard scale made it possible to combine all 30 indicators and calculate an accurate final score.