A year ago, I wrote two posts looking at the Covid-19 numbers for Canada.

• Part I looked at a basic exponential model built at a time (mid-March 2020) when we only had 350 reported cases across the country. It predicted that we'd have 26,000 two weeks later, which sounded absolutely crazy. It was wrong by about two weeks. We hit 26,000 in mid-April 2020.

• Part II (beginning of April 2020) looked at different logistics models that we could fall on depending on how hard we worked at flattening the curve. As of today, we have 927,069 cases country-wide so that didn't happen.

Instead, this is what happened:

• We can think of it as a logistics curve that plateaued at a little over 100,000 cases around June 2020.
• But then in September 2020 it kicked into exponential gears again (the second wave).
• It looked like it was starting to plateau again in mid-January 2021, but instead, we might be headed into a third wave.

I'm now keeping a public spreadsheet analyzing all of this where I use four different graphs to gain insights into where things are going.

This graph is a “smoothed out” version of the actual reported counts. Testing backlogs that get caught up later creates noise in the original data that I reduced using a 7-day running average, which seemed to be a good balance between details and noise level.

This graph represents the number of new cases each day. Mathematically, it's the first derivative of the previous graph. Numerically, I computed the slope of 5-day secants of the smoothed curve, again to reduce on the overall trend.

Here, we clearly see that from June to September were our lowest number of new cases, which translated into an almost horizontal curve above. And we can clearly see the “height” of the second wave in the winter, and how we might be about to start a third wave now without even having finished the second one.