Written by Bertil Helseth
on March 26, 2020

DISCLAIMER: The modeling effort described in this article aimed to predict the first wave of COVID-19 related deaths and is not updated since May 3, 2020. The graphs and explanatory text will not be updated.


At Intelecy we deal with data, predictions and forecast every day.

Our team decided to put aside other work for one day and see if we were able to get some better insight to the COVID-19 pandemic. We created an interactive analysis that can show, compare, and project possible answers to the big questions surrounding this pandemic. We will keep this analysis updated with the most recent data in the weeks to come. Our goal is to create an interactive analysis that can both show, compare and project on the big questions surrounding this pandemic, updated with the most recent data.

  • What is likely to happen next with COVID-19?

  • Are things getting better or worse right now?

  • When will we reach the peak?

  • When will things return to normal?

We wish to emphasize that we don’t know what will happen and there is still a lot of uncertainty both in actual data we are using and measures taken to slow down the spread, so our estimates are not to be taken as any conclusion.

Intelecy has made all the analysis and code available to anyone who would like to investigate further. This article is all about the results of our analysis. Interested in the deeper technical details of our data, methods, and models?

Links are at the bottom of this page.


Relative Velocity of COVID-19

Interactive plots: Click the graphs below to use the interactive plots. You can zoom by selecting the area. Double click will take you back to default zoom. Single click on the labels (countries) will select/deselect, double click on labels will show only the selected country. You can see a quick demo of this interaction in the animation at the top of the article.

Below you can test the interactive charts for yourself.

These two charts compare the progress of the COVID-19 virus on a country-by-country basis. First in terms of cumulative confirmed cases, second in terms of cumulative deaths. We show historical data, where the start time is fixed to 5 confirmed deaths.



Predicted COVID-19 Body Count

The two next charts show both actual deaths and our projections in a combined view. The Cumulative Deaths Forecast shows the total number of deaths per country, while the Daily Deaths Forecast predicts the peak of the curve, answering the question: When will things start to get better?

If you are curious about our modeling and considerations behind these forecasts, please scroll down to read more about our Modelling and Considerations behind these predictions.plot3plot4


In our analysis we have used the COVID-19 data from the Johns Hopkins Coronavirus Resource Center and population estimates from the United Nations.

Our analysis focuses on modelling the reported deaths due to the virus. The number of confirmed cases would give us more data but is highly dependent on the amount of testing available, the quality of testing, and the level of reporting in any given country. For this reason, we feel that the reported deaths total give us a better baseline for comparisons across countries. We have limited the analysis to countries that have more than 1000 confirmed cases and a minimum of 5 deaths.


Modelling and considerations

We are modelling the spread of the virus as a logistic function. This function is used to model growth in a constrained environment, where the growth rate will reach a peak then start to slow down. It is commonly used in biology to model population growth and also appears to be a good fit for the progression of COVID-19.

We then map multiple possible logistic functions onto the existing data to give a distribution of possible outcomes.

For countries where the virus is further progressed, for example (e.g. China and Italy), the possible outcomes are tightly grouped together. However, for countries early in the outbreak, (e.g. the United States) there is a significant variation in the possible outcomes, showing a much larger uncertainty.


These predictions are the result of our best efforts as a team. However, there are several factors that affect the accuracy of the forecasts. Some countries, including Norway, have had very few COVID-19 related deaths. Which is good! But limited data means limited modeling capability. Other countries, such as the United States, are quite early in the spread of the virus, making it much harder to accurately predict outcomes. Read more about our technical considerations in the technical blog post.


Resources and call to action

We decided to open source all the code and write a technical blog about the methods and modelling we have done. For anyone interesting in contributing and improving our work you'll find links to the resources below.

If you have good suggestions or otherwise would like to get in touch with us, send emails to


Part 2: Technical Blog post about the COVID-19 analysis

Link to Colab

Link to Github

You may also like:


Invests in the next phase of autonomous industrial production

When the Asian markets slowed down because of the COVID-19 situation, Intelecy was quick to accelerate the plans to focu...

Press Blog

COVID-19 Probabilistic inference with logistic growth

Updated: Sep 30, 2020 Authors: Are Haartveit, Harald Husum and Tom Van de Wiele with equal contributions. Continuation o...


Ambassador for Norwegian technology on the world market

“Intelecy is a good example of what works. This is a Norwegian quality solution that scales worldwide and where we are p...