Methodology

We collected Monthly Consumer Price Index (CPI) data for 17 major cities in Pakistan from January 2023 to January 2025. The raw data was parsed from Pakistan Bureau of Statistics (PBS) reports.

• 7 Categories: Food, Utilities, Transport, etc.
• 17 Cities: Including Islamabad, Lahore, Karachi, Quetta, Peshawar.
• Normalization: Prices were normalized to a [0, 1] scale to allow comparison across different item types.

We treat each city as a node in a graph. An edge exists between two cities if their price trends are highly similar. We used Cosine Similarity to measure this relationship:

To identify economic hubs, we calculated four key centrality metrics for every city in the network:

• Degree Centrality: The number of direct connections a city has. Indicates immediate influence.
• Closeness Centrality: The inverse of the average shortest path distance to all other cities. Indicates how quickly price shocks in this city can spread to others.
• Betweenness Centrality: The fraction of shortest paths that pass through a city. Indicates a city's role as a bridge or broker between different regions.
• Eigenvector Centrality: A measure of influence where connections to high-scoring nodes contribute more to the score. Indicates connection to other powerful hubs.

To produce a single ranking, we aggregated these centrality scores using two weighting schemes:

• Equal Weights: All four centralities contribute equally (25% each).
• Entropy Weights: Weights are determined by the information entropy of each metric. Metrics with higher variability (more information) receive higher weights.

Finally, category-specific scores were aggregated into a Global Influence Score using economic importance weights (e.g., Food Staples have a higher weight than Clothing).