Here are some Python Jupyter data science notebooks from October 2020 analyzing world COVID-19 data available from Our World in Data.
They are some of (the better) solutions produced by students as the first mini-project (out of three) in P176M010 Advanced Machine Learning course at Kaunas University of Technology Faculty of Informatics given by Mantas Lukoševičius.
Students had to come up with their own unique questions and answer them using the data. A solution template notebook was provided. A snapshot of the https://covid.ourworldindata.org/data/owid-covid-data.csv data from that time is included. Some of the solutions use additional sources of data.
The notebooks are provided for reference and instruction with no warranty of any kind. They are in the state that they were submitted originally. Some additional data needed to run them might not be easily shareable. For that you may contact the authors directly.
- How did the value of 10 Lithuanian stocks change on the first day of quarantine?
- Which stocks were most affected during quarantine?
- Do the COVID cases help predict future stock values?
- Are more tests performed in countries with higher GDP per capita?
- Does higher diabetes prevalence implies that the country has higher morbidity rate among infected people?
- Is the lockdown necessary in the Czech Republic? (Analyze how it helped last time, predict the number of cases if no new restrictions will be announced..)
- What is the average % of people who wear a mask in public places for each continent today?
- What is the correlation between a country having a mask-wearing culture history and amount of COVID-19 cases?
- What are the predictions of COVID-19 cases for countries which started wearing masks versus countries which did not adopt mask wearing?
https://today.yougov.com/topics/international/articles-reports/2020/03/17/personal-measures-taken-avoid-covid-19 (Wearing a face mask when in public places)
- Which country was leading in testing from January to June 2020?
- Correlation between total tests and total cases in Lithuania, Latvia and Estonia from January to June 2020?
- Can we predict how many new cases Lithuania, Latvia and Estonia will have in November 2020?
- Which European countries did best and worst to "flatten" the active cases curve?
- Is COVID-19 case fatality rate lower in European countries that have a flatter active cases curve?
- Does using active cases data improve prediction accuracy of daily deaths in Belgium compared to using only historical daily deaths data when using Linear Regression?
https://github.com/CSSEGISandData/COVID-19
- How do highly libertarian vs. highly authoritarian governed countries perform wrt "wave mitigation"?
- Can the temporal delay between an increase in Covid cases and deaths be quantified by utilizing Dynamic Time Warping?
- Trying to short-term forecast the amount of daily new Covid cases using k-nearest-neighbors retrieval and adaptation (case-based approach).
https://freedomhouse.org/report/freedom-world
- Which country had the biggest death increase per 1 000 000 inhabitants each month?
- What is the correlation between numbers of cases in the Slovak Republic and a number of cases in the Czech Republic?
- Would be possible to predict free movements between Slovak and Czech republic by using Classifier Decision TreeClassifier, RandomForestClassifier and AdaBoostClassifier
Own dataset
The course was aided by TAs: Lukas Stankevičius and Eglė Butkevičiūtė.
The authors agreed to their work being published under a Creative Commons Attribution 4.0 International license.
So is this intro summary by Mantas Lukoševičius.