CXC 2025 - SAP Challenge

Inspiration

Understanding poverty requires more than just measuring income. Our inspiration was to create a comprehensive, data-driven tool that identifies the root causes of poverty in different countries. By using multidimensional indicators and advanced analytics like PCA weighting, we aimed to deliver insights that can guide targeted policy interventions and support the United Nations Sustainable Development Goals.

What it does

Our project calculates a Multidimensional Poverty Index (MPI) for each country using key indicators such as:

Education: Literacy rate, school enrollment
Health: Infant mortality rate, access to healthcare
Living Standards: Access to electricity, clean cooking fuels, and safe drinking water
Income: GDP per capita and employment rates

Using Principal Component Analysis (PCA), we assign weights to each indicator based on its contribution to overall poverty. The final MPI scores are visualized through:

Bar charts highlighting countries with the highest and lowest MPI scores
An interactive choropleth map that allows users to explore global poverty trends
A dashboard combining heatmaps, time-series graphs, and country comparisons

How we built it

Data Cleaning & Preprocessing:

Handled missing data using row-wise mean imputation
Normalized indicators using Min-Max scaling (0-1 range)
Filtered out rows with more than 90% missing data

PCA Weighting:

Applied Principal Component Analysis (PCA) using Scikit-Learn
Used the first principal component (PC1) to assign weights based on variance
Combined indicators into a weighted MPI score for each country

Visualizations:

Plotly Express: Created interactive bar charts and an interactive choropleth map
Matplotlib and Seaborn: Developed heatmaps and scatter plots for analysis

Challenges we ran into

Data Quality: Managing datasets with missing values and inconsistent formats was time-consuming
Normalization: Ensuring that indicators with different scales were fairly compared

Accomplishments that we're proud of

Created a visually engaging interactive choropleth map
Integrated bar charts, heatmaps, and time-series graphs to support our insights
Simplified the presentation to clearly communicate our methodology and findings

What we learned

Data cleaning and normalization are essential to ensure fair comparisons between countries
Interactive visualizations are crucial for communicating data insights