How We Used Alteryx to look at Covid-19 Data Driven Decisions

As I write this post, we’re in the middle of an unprecedented public health crisis. Twitter is alive with self-proclaimed experts variously recommending preparation for the End Times or simply ‘taking it on the chin’.

One of the really empowering things about using Alteryx is that everyone can become a citizen data scientist. This week, we’ve seen an exponential curve kicking off in Italy, and I really wanted to know what a time series model would look like for the UK data.

[Caveat emptor: this should not be taken as an authoritative work, nor should the reader make any decisions about his/her future based on my analysis]

So, this can be achieved in three easy steps.

1) Download the EU’s Centre for Diseases Control’s summary of the Covid-19 data from here.

2) Clean up the data ready for creating an ARIMA forecast.

3) Drag the ARIMA tool on to your workflow.

Out of the box, Alteryx gives a dynamic ARIMA graph that allows you to choose a date in the future, the dotted lines indicate lower confidence levels. The exponential curve for Italy shown below, indicates why we should all take the spread of the virus seriously, and that delaying basic social distancing strategies is inexcusable.

Using the Time Series tool, Alteryx allows you to extract the raw ARIMA predictive data, I then wrote a batch macro, to batch through each country, and sent the finished data to Tableau. The Tableau output allows the user to visually compare the progress (and predicted progress) of the disease country by country.

I am not a data scientist, but the predicted level off in Chinese infection rates versus the exponential curves faced by Western countries that are slow to adopt containment strategies, is self-evident.