INTRODUCTION

This COVID-19 Data Analysis project aims to analyse the impact of the pandemic on global health using SQL. The project will utilise publicly available datasets to extract insights into the spread and severity of the virus.

ABOUT THE DATASET

For this project, I am using the Coronavirus (COVID-19) Deaths dataset, published on Our World in Data under the Creative Commons BY license. This dataset contains a daily log of Covid-19 cases and other Covid-19-related statistics and has been updated every day since 8 January 2020.

Citation
Edouard Mathieu, Hannah Ritchie, Lucas Rodés-Guirao, Cameron Appel, Charlie Giattino, Joe Hasell, Bobbie Macdonald, Saloni Dattani, Diana Beltekian, Esteban Ortiz-Ospina and Max Roser (2020) – “Coronavirus Pandemic (COVID-19)”. Published online at OurWorldInData.org. Retrieved from: ‘https://ourworldindata.org/coronavirus’ [Online Resource]

CHARACTERISTICS AND POINTS OF INTEREST

This dataset contains several attributes, not all of which will be used for this analysis. Data can be broken down into the following categories.

For the purpose of this project, I will only be using the Case-load and vaccination data, as this will only be relevant for this initial study.

---

DATA EXPLORATION

The data exploration phase involves an unstructured examination of a large dataset to identify initial patterns, characteristics, and points of interest.

During this phase, I explore the data without a specific plan or hypothesis, aiming to gain a broad understanding of the dataset before proceeding to more structured analysis techniques.

I’ve provided my exploratory SQL queries here for reference.

INITIAL OBSERVATIONS

The dataset contains daily reports of COVID-19 cases and deaths across multiple countries. The data spans back to the initial detection of the virus in each country, providing a comprehensive overview of the pandemic’s spread.

One notable feature of this dataset is its cleanliness, with no instances of outdated, duplicated, or incomplete data. However, some entries deviate from the standard location listing format by specifying the location as “World” or a continent name. This poses a challenge as it could lead to skewed comparisons with country-specific data. Addressing this issue is crucial in the data-cleaning phase.

Moreover, some NULL values are present alongside continent-specific data entries. It is essential to filter out these values during the cleaning phase to ensure the accuracy of subsequent analyses.

Follow along on as I continue with the data cleaning phase of the project, where the focus will be on identifying and eliminating any errors, inaccuracies, or inconsistencies that could impact the reliability of the data. This meticulous approach will ensure the dataset is robust and reliable, setting the foundation for insightful analysis.

---

DATA CLEANING LOG

Filtering out the aggregate rows 

The creators of the dataset have gone the extra mile by including rows for Africa, Asia, Europe, the European Union, North America, Oceania, and South America. While this is helpful for easy reference, it could potentially distort the analysis, as these are aggregate figures representing all the countries within those continents. As such, I have decided to filter out those rows to allow for a more accurate comparison of country-specific data.

Filtering out income categories from the Location column 

Furthermore, the dataset contains new income parameters that are irrelevant to our use case, namely, low-income, lower-middle income, upper-middle income, and high-income. To prevent any potential interference, I have filtered out these rows.

Filtering out locations that form part of one country

A few location-based reports feature rows for separate territories within a country. For example, England, Northern Ireland, Wales, and Scotland have all been recorded separately in the ‘Location’ field, despite being part of the United Kingdom. As no data has been reported against these locations, I have filtered out these rows.

Verifying the country-specific data and checking for duplicates 

There are 195 countries, and I can see the reported data for those countries at the top of the list. However, there are a total of 249 rows. 

Upon inspection of the dataset, I discovered a few instances where the continent has been listed as a country. I also identified a few territories that have been named incorrectly, such as the United Kingdom. 

Additionally, some territories that are not officially recognised as countries have reported Covid-19 data to the World Health Organisation. After careful review, I found that there were no duplicates, and I was left with a final list of 195 countries.

Filtering out countries and territories with NULL values 

Some regions had NULL values, which must be filtered out before I proceed with the visualisation. These included Taiwan, North Korea, Hong Kong, Turkmenistan, Macao, Northern Cyprus, and Western Sahara.

Changed date format 

Finally, I decided to standardise the format of the sale date by converting it from a date-time format to a Year-Month-Day format. This change makes the data easier to interpret and allows for better trend analysis over time.

---

DATA VISUALISATION WITH TABLEAU

---

CONCLUSION

Understanding the data surrounding the Covid-19 pandemic is crucial for grasping its comprehensive impact. Even post-pandemic, analysing this data provides invaluable insights for future outbreak preparedness, informing public health strategies and enhancing global resilience against similar threats.

---

Thank you for reading this project.

If you’d like to share your thoughts or feedback, please feel free to reach out.