Spatial Analysis of COVID-19 Infection Patterns Using Unsupervised Classification with K-Means Clustering

Although there has been extensive research on COVID-19 spread, however, a notable gap continues to exist in comprehending the spatial variability of its spread through advanced geospatial techniques. Many studies focus on temporal trends or analyses at the national level, frequently overlooking localized differences that affect the progression and severity of diseases. In this study, we conducted an in-depth Exploratory Data Analysis (EDA) to examine the global spatial and temporal distribution of COVID-19 cases. Using visual analytics such as count plots, bar charts, histograms, and box plots, we identified key trends in confirmed, active, recovered, and death cases. Our findings show that South Africa had the highest confirmed cases in Africa, while Russia and the United Kingdom led in Europe. Iran, Pakistan, and Saudi Arabia reported the highest cases in the Eastern Mediterranean, whereas the United States and Brazil recorded the highest peaks in the Americas. India had the most cases in Asia, while China reported the highest case counts in the Western Pacific. A correlation matrix indicated strong positive relationships between confirmed cases and deaths (0.91), recoveries (0.90), and active cases (0.95), showing that higher case numbers significantly influenced mortality and recovery trends. Spatial dependence analysis using Moran’s Global Index and the Getis-Ord Gi statistic confirmed notable clustering of COVID-19 cases. Hotspot analysis identified high case concentrations in North America, Europe, and Asia, while Africa and parts of South America exhibited lower infection rates. K-means clustering categorized the pandemic’s spread into two distinct clusters regions with the highest and lowest case counts. Our findings emphasize the non-random geographic distribution of COVID-19 cases, highlighting regional disparities and providing insights for targeted interventions.