- Identify Hospital Dataset: Locate a dataset with comprehensive information on hospitals across India.
- Extract and Filter Relevant Data: Extract necessary information (e.g., hospital names) and filter out irrelevant values.
- Web Scraping Setup: Develop and test a script to scrape data (coordinates, ratings, number of ratings) from Google Maps for each hospital.
- Data Documentation: Document scraped data, including coordinates, ratings, and review counts.
- Define Effective Rating Formula: Implement a formula combining ratings and review counts to calculate an effective rating.
- Determine Sphere of Influence: Develop an algorithm to calculate influence radius based on each hospital's effective rating.
- Grid the Country: Create a 1 km² grid overlay for India.
- Get Census Data: Devise a means to use census data for the project
- Calculate Influence Per Cell: Sum influence scores from each hospital within the radius of each cell, and multiply by census data for population density.
- Define Low-Access Thresholds: Establish thresholds to classify cells as “low-access” based on influence scores.
- Visualize Low-Access Areas: Use mapping libraries to create a visual representation of underserved areas.
- Define Anonymization Techniques: Select appropriate anonymization techniques for hospital ratings and coordinates.
- Test Cluster Consistency Post-Anonymization: Re-apply clustering to anonymized data and compare results with original clusters.
- Document Anonymization Methodology: Publish a clear guide on anonymization techniques that maintain data utility.
- Prepare Final Report and Visualizations: Summarize findings and create visualizations of influence distribution and underserved areas.
- Effective Rating: Create a formula that combines the rating (1–5 scale) and the number of ratings to balance quality and popularity. For example:
- Effective Rating: Create a formula that combines the rating (1–5 scale) and the number of ratings to balance quality and popularity. For example:
Effective Rating = Rating * log(1 + Number of Ratings) - Sphere of Influence: Define the influence radius using an exponential decay function:
where
Radius = R0 * exp(alpha * Effective Rating)R0is a base radius, andalphais a factor controlling the decay rate.
- Overlay a grid over the country (e.g., each cell representing 1 km²).
- For each hospital, calculate its influence within each cell inside its radius, then sum influences from all hospitals at each cell.
- Multiply the influence value for each cell by its population density (using census data) to highlight underserved areas.
- Define thresholds to identify “low-access” areas based on influence scores (e.g., bottom quartile).
- Visualize these areas on a map to identify patterns, especially in rural or densely populated areas.
- Coordinates: Apply slight perturbations within a tolerance that doesn’t significantly impact proximity-based influence.
- Effective Rating: Implement generalization or value swapping within a defined range so that ratings maintain influence while masking original values.
- Ensure the anonymization process preserves the distribution and relative influence of each hospital.
- Document methods to replicate the anonymization process while ensuring data utility.
- Demonstrate consistency between anonymized and original data’s influence maps, ensuring accuracy for public sharing.