Updated: Sep 12, 2020
A couple months ago, I was quite perturbed by the pandemic, as I am sure everyone else was. So to try to better understand the situation, I took to the task of visualising it differently. We all have seen new cases along time or total cases against time:
However what was missing was if the curve was the type of trend: exponential, polynomial or something different. In addition, these graphs made it hard to make predictions.
To solve this problem I coded a python program that calculates new cases / total cases which by some math property shows the type of acceleration and in some cases allows for predictions. Below is an example of such graph:
From this we can see a clear linear trend. The implementation of such analysis was quite straightforward. The difficulty was sourcing the data, luckily I found a Kaggle source which redirected me here.
From this graph we can also see the type of covid growth in Brazil. I would describe it as exponentially decreasing new cases which is good! From this graph, what would also be possible, you could do a linear fit and calculate the natural curve of the total cases function.
Thank you for reading this post. I hope you learnt something as I did. If you are interested in the source code or the graphs for all the other countries of the world, you can check out this github here.
Also feel free to use the code to create your own prediction algorithms and remember to post about the results here!