WHO COVID-19 Global Table Data
- Rows
- 19
- Columns
- 12
- Data range
- American Samoa to Suriname
- Auto-refresh
- Re-checks its source every 30 days
- Last refreshed with new data
- 2026-10-13
Global COVID-19 cases and deaths table data. Time-series from WHO.
AI analysis
WHO COVID-19 Global Table Data — Analytical Assessment
1. Dataset Overview and Correlation Value
This dataset represents a WHO-published global COVID-19 summary table, delivered as a time-series snapshot aggregated at the regional or country level across 19 rows and 12 columns. Sourced from the WHO's Azure blob storage endpoint (srhdpeuwpubsa.blob.core.windows.net/whdh/COVID/WHO-COVID-19-global-table-data.csv), the publisher is authoritative — the World Health Organization — lending strong credibility to the cumulative figures, though the blob URL structure suggests this may be a static or infrequently refreshed export rather than a live feed, which has implications for temporal freshness. With both raw counts and per-100,000-population normalized metrics present simultaneously, this dataset is particularly well-suited for cross-country correlation studies where population size would otherwise confound comparisons. The co-presence of cumulative totals and recent reporting windows (7-day and 24-hour columns) theoretically enables both longitudinal burden analysis and near-real-time trend correlation, though as noted below, the recent-activity columns are severely compromised.
---
2. Data Quality Observations
Data quality is a significant concern for any analytical work on this dataset. Across 12 columns and 19 rows, there are 103 total null cells, representing roughly 45% of the entire data matrix — a striking rate that warrants caution before any modeling or correlation work proceeds. Two columns are completely empty: Cases - newly reported in last 7 days per 100000 population (100% null, 19/19) and Deaths - newly reported in last 7 days per 100000 population (100% null, 19/19), rendering them analytically useless in their current state and suggesting either a data pipeline failure or a deliberate deprecation of those fields. Additionally, Cases - newly reported in last 7 days and Cases - newly reported in last 24 hours each carry 16 nulls (84.2%), leaving only 3 usable rows for recent-activity analysis — far too sparse for reliable correlation. A notable type mismatch also exists: Deaths - newly reported in last 7 days per 100000 population is typed as String despite representing what should be a numeric rate, suggesting either mixed-type raw data or a serialization error at ingestion. There is also one null in WHO Region, which could disrupt any region-level aggregation or grouping. Duplicate row counts are pending Phase D recomputation and should be verified before any deduplication-sensitive joins.
---
3. Key Column Distributions
The most analytically rich columns are the cumulative totals and their per-100,000 normalizations. Cases - cumulative total spans from 8,359 to ~99.4 million (mean: ~5.91M, median: ~88,953), with a dramatic skewness of 4.31 and 2 outliers — almost certainly corresponding to the highest-burden nations (likely the US or India at the top end), pulling the mean far above the median and making raw totals unsuitable for direct comparison without normalization. Cases - cumulative total per 100000 population tells a somewhat more balanced story (min: 40, max: 51,256, mean: 12,703, median: 8,250, skew: 1.55, 2 outliers), though still right-skewed, indicating that even on a per-capita basis, a small number of regions experienced dramatically higher case burdens. Deaths - cumulative total mirrors this pattern with a skew of 4.07 and 2 outliers (min: 34, max: 122,398, mean: ~9,992, median: 1,024), while Deaths - cumulative total per 100000 population is more tractable (min: 1, max: 406, mean: 86, median: 62, skew: 1.95, 1 outlier). The recent-activity columns (Cases - newly reported in last 7 days, Cases/Deaths - newly reported in last 24 hours) show near-zero variance with means of ~1.33 or 0 and are effectively binary or zero-inflated given the sparse data — these should be treated as unreliable for any analytical purpose in the current dataset snapshot.
---
4. Recommended Join Key Columns
The Name column is the natural and strongest join key, confirmed by its 0 nulls and 19 fully distinct values, making it a clean unique identifier for every row (likely country or regional entity names). WHO Region serves as a secondary grouping key for region-level joins, though its single null value requires imputation or exclusion before use. For cross-dataset joins, it is strongly recommended to standardize Name values against ISO 3166-1 country codes or UN M49 region codes, as WHO naming conventions can differ from other datasets (e.g., "United States of America" vs. "USA" vs. "US"). Any joined dataset should carry either a matching name field or an ISO code that can be mapped to WHO names via a lookup table. The per-100,000 columns are ideal join targets (rather than keys) for regression or correlation analysis, as they are already population-normalized.
---
5. Recommended Complementary Datasets for Correlation Discovery
Several dataset categories would pair powerfully with this data:
- Vaccination coverage data (e.g., WHO/Our World in Data vaccination rates by country) — to correlate cumulative case/death burdens with vaccination uptake percentages, potentially revealing dose-response relationships at the population level. - Healthcare system capacity indicators (e.g., World Bank hospital beds per 1,000, ICU capacity, physician density) — to test whether health infrastructure correlates with per-capita death rates, controlling for case load. - Socioeconomic indices (e.g., World Bank GDP per capita, Human Development Index, Gini coefficient) — to examine whether economic inequality or national wealth correlates with case detection rates or mortality outcomes. - Demographic datasets (e.g., UN Population Division age structure data) — given COVID-19's well-documented age-skewed mortality, median population age and the proportion of population over 65 would be strong correlation candidates against death-per-100,000 figures. - Non-pharmaceutical intervention (NPI) stringency data (e.g., Oxford COVID-19 Government Response Tracker / OxCGRT) — to correlate lockdown stringency or policy response timing with cumulative case trajectories across the same WHO-named entities.
Columns
- Name (string)
- WHO Region (string)
- Cases - cumulative total (integer)
- Cases - cumulative total per 100000 population (integer)
- Cases - newly reported in last 7 days (integer)
- Cases - newly reported in last 7 days per 100000 population (integer)
- Cases - newly reported in last 24 hours (integer)
- Deaths - cumulative total (integer)
- Deaths - cumulative total per 100000 population (integer)
- Deaths - newly reported in last 7 days (integer)
- Deaths - newly reported in last 7 days per 100000 population (string)
- Deaths - newly reported in last 24 hours (integer)