Exploratory Data Analysis: WHO COVID-19 data

Introduction

This report outlines the process followed during an Exploratory Data Analysis (EDA) on data published by the World Health Organisation (WHO) data on daily COVID-19 cases and deaths by date reported to the WHO. ¹

The dataset (as retrieved on 2022-12-12) contains 254064 records (this dataset continues to be updated and will contain more records if retrieved at a later date), each with 8 attributes:

Name	Type	Description
Date_reported	Date	Date of reporting to WHO
Country_code	String	ISO Alpha-2 country code
Country	String	Country, territory, area
WHO_region	String	WHO region offices
New_cases	Integer	New confirmed cases
Cumulative_cases	Integer	Cumulative confirmed cases reported to the WHO to date
New_deaths	Integer	New confirmed deaths
Cumulative_deaths	Integer	Cumulative confirmed deaths reported to the WHO to date

Missing Data

Null values

The dataset has no null values in any column, with the exception of the Country_code column, which is missing 1072 values (0.42% of rows in the dataset).

Upon inspecting the rows with missing data, it seems that only records with Country values of “Other” are missing. The WHO state that this category indicates:

Other: cases and deaths reported from international conveyances, included in global totals but not reflected in epidemiological curves as not associated with a country or region.

This data has been retained, in order to avoid incorrectly adjusting the global figures on COVID-19 cases and deaths.

That being said, the data with no ISO Alpha-2 country code ² has not been mapped in the chloropleth maps shown below, as these country codes are the means by which the COVID-19 data from the WHO was merged with the geographical representation of each country’s boundary, and with its estimated population (both of which are available in a single GeoJSON file from the World Bank). ³

Data not recorded

Although – as described above – there is very little missing data in the dataset in terms of null values, there is likely to be a significant amount of data which is completely missing from the dataset (having never been recorded and reported to the WHO).

Cases of COVID-19 which have gone untested and undiagnosed – and deaths resulting from cases of COVID-19 which were not detected – will not have been reported to the WHO.

There are many possible causes for failure to identify COVID-19 cases and to report them to the WHO. These may include, but are not limited to:

• shortages of testing kits and facilities
• lack of access to medical care of any kind
• lack of access to communications networks (e.g., the internet) and therefore to avenues for reporting positive test results
• deliberate avoidance of testing in order to avoid local, national and international movement restrictions, or due to a financial imperative to stay in work whilst infected
• failure to self-test due to a lack of concern about the implications of having COVID-19 (especially following widespread vaccinations and easing of restrictions)
• failure of national agencies to report cases to the WHO for political reasons

The data presented below should be viewed with this limitation in mind.

Data Cleaning

The data supplied by the WHO was already very clean and complete (with the caveat regarding unreported data as described above).

The values in the Date_reported column were parsed to python datetime objects (from string representations of dates as provided in the original dataset) using pandas' to_datetime method in order to facilitate easy manipulation of the date data.

Data Exploration

Investigation of COVID-19 cases and deaths by WHO region

The WHO divides the world into six regions, and has a regional office for each of these regions. These are:

WHO region code	Description
AFRO	Regional Office for Africa
AMRO	Regional Office for the Americas
SEARO	Regional Office for South-East Asia
EURO	Regional Office for Europe
EMRO	Regional Office for the Eastern Mediterranean
WPRO	Regional Office for the Western Pacific

Figure 1 shows the total number of (all-time) COVID-19 cases reported to the WHO by region. The data indicates that Europe and the Americas have recorded the largest numbers of cases of COVID-19 since the beginning of the global pandemic, whilst the Eastern Mediterranean and Africa have recorded the fewest.

Figure 2 shows the total number of (all-time) COVID-19 deaths reported to the WHO by region. The number of deaths in each region correlates as expected with the number of cases in each region – larger numbers of cases predict larger numbers of deaths.

Note that the order of the regions is not exactly the same as that shown in Figure 1. For example, whilst Europe recorded higher case numbers than the Americas, the Americas recorded a larger number of deaths. This may reflect differences in the overall standard of healthcare provision between regions (for example, that healthcare provision in Europe is better in general than healthcare in the Americas – or at least that treatment of COVID-19 cases was more effective in Europe).

Figure 3 shows the number of COVID-19 deaths per 100 cases reported to the WHO by region. The result shown is interesting in that, whilst (as shown in Figures 1 and 2) the WHO’s Africa region reported the lowest numbers of cases and deaths, it is the region with the highest death rate for confirmed COVID-19 cases.

This may indicate several things. One interpretation of this result is that access to testing and healthcare facilities in the Africa region was lower than in other regions, meaning that both more confirmed cases of COVID-19 resulted in death, but also that many more mild cases of COVID-19 went untested and undiagnosed and therefore were not reported to the WHO (so are not present in the dataset), thus disproportionately weighting the Africa region’s data with more serious – and fatal – cases and artificially inflating the number of deaths per 100 cases relative to the other regions.

Figure 4 shows a scatter plot of total (all-time) COVID-19 deaths against total (all-time) COVID-19 cases by country. The best-fit line indicates the positive correlation between numbers of cases and numbers of deaths noted above (see Figures 1 and 2) more clearly. Note that the international data noted by a country value of “Other” has been excluded.

Global COVID-19 cases and deaths over time

Figure 5 shows the cumulative totals of COVID-19 cases and deaths reported to the WHO over the course of the pandemic. Whilst numbers of cases are in the order of hundreds of millions, deaths are in the order of millions. At a glance, this tells us that the global death rate for confirmed cases of COVID-19 is around 1% (1.29% to be precise).

As described above, it is likely that the true death rate is lower than this (as many mild cases of COVID-19 will not have been confirmed and reported to the WHO).

We can also see a significant increase in the rate of increase of global cumulative case numbers in January 2022 (with a less pronounced increase in the rate of increase of deaths lagging behind by about a month), which had been increasing nearly linearly throughout 2021.

Figure 6 shows the numbers of new confirmed COVID-19 cases and deaths reported to the WHO each month over the course of the pandemic. Data for December 2022 is partial and has therefore been excluded to avoid confusion.

The spike in global case numbers in January 2022 described above can be seen more clearly in this visualisation. We can also see the positive impact of the global vaccination program in the lower numbers of deaths per month in late 2021 and 2022 (despite similar or higher numbers of cases per month).

UK COVID-19 cases and deaths over time

Figure 7 shows the cumulative totals of COVID-19 cases and deaths in the UK reported to the WHO over the course of the pandemic. This data at a national level is much less smooth than the global data (which is likely smoothed by the local spikes in cases and deaths happening at slightly different times).

Figure 8 shows the numbers of new confirmed COVID-19 cases and deaths in the UK reported to the WHO each month over the course of the pandemic. Data for December 2022 is partial and has therefore been excluded to avoid confusion.

The impact of the vaccination program is even more pronounced here – with spikes in case numbers corresponding to significant spikes in deaths in 2020 and early 2021, and far fewer deaths resulting from the larger spike in cases in winter 2021-2022.

A visualisation like this one could be compared with the dates of the national lockdowns⁴ to better understand the reasoning behind those lockdowns, and to evaluate their effectiveness.

Investigation of COVID-19 cases and deaths per 100,000 population by country

Figure 9 shows the number of COVID-19 cases per 100,000 population for each country in the WHO dataset.

As noted in the Missing Data section, the figures not associated with a particular country have been excluded from the maps below.

Western European countries immediately stand out as having recorded very high numbers of cases relative to their populations.

Figure 10 shows the number of COVID-19 deaths per 100,000 population for each country in the WHO dataset.

Peru immediately jumps out as having been particularly badly affected in terms of COVID-19 deaths relative to its population. This result was discussed by the BBC in June 2021. ⁵

---

1. WHO, WHO Coronavirus (COVID-19) Dashboard, accessed 2022-12-12 ↩︎

2. GOV.UK, ISO Country Codes, accessed 2022-12-15 ↩︎

3. World Bank, World Bank Official Boundaries, accessed 2022-12-13 ↩︎

4. Institute for Government, Timeline of UK government coronavirus lockdowns and restrictions, accessed 2022-12-21 ↩︎

5. BBC, Covid: Why has Peru been so badly hit?, accessed 2022-12-21 ↩︎