You can add location information to your Tweets, such as your city or precise location, from the web and via third-party applications. You always have the option to delete your Tweet location history. Learn more
What do you think? • Rolling average, because it irons out false precision from noisy daily reporting and highlights medium-term trends? • Spline, because we still have smoothness for following individual lines, but we’re also more faithful to that day’s reported data point?
And now the other charts. First, the spline method for new daily cases: • New confirmed US cases may have peaked, same for UK. Does this mean outbreaks have peaked? Time will tell • Austria cases trending down; they plan to start easing lockdown next weekpic.twitter.com/zZ0y0EI29c
Now regions: • The spline still sees New York’s daily deaths trending upwards every day, same for London • New Jersey announced record new death toll today, sending its curve bending back up • Washington state now starting to trend up long after initial casespic.twitter.com/lJBaaQVyHI
And finally small multiples for daily deaths for 47 countries:pic.twitter.com/tCMTQCxwi3
Hopefully this also gives some insight into the huge amount of work that goes into these behind the scenes, the amount of thought that goes into every pixel, and the constant compromises that must be made. The charts you see every night are just the last step in a giant process.
I’d also add that digging into the innards of this today, and playing more attention to the raw daily numbers for smaller countries, has left me very concerned about data quality. Vast amounts of the data coming in from individual countries is essentially junk.
Take Ecuador, where according to the data, deaths are trending downwards into single digits, yet literal vultures are circling overhead as coffins lie in the streetshttps://www.ft.com/content/5e970473-0710-44f6-bfae-2a830b78a3a1 …
This is one of countless examples. Look at the lines arcing downwards in the lower left. I’m skeptical that more than one or two are truly containing the outbreak. Data quality is a fundamental problem, but one advantage of the rolling avg is it irons out those daily quirks.pic.twitter.com/69NLwqNneE
Far more likely is that lower curves for developing countries represent massive under-testing, and an overwhelmed health statistics infrastructure. Even in developed countries we know the stats are flawed, but this will be worse elsewhere:https://www.ft.com/content/c07e267b-7bca-418f-ad9e-8631a29854cb …
So maybe the rolling average will return tomorrow, as it removes the illusion that daily numbers really represent any kind of "ground truth", and better represents what they *do* show: A foggy picture of general trends in messy outbreaks that are incredibly difficult to track.
I’m still glad I did the splines, even if it did keep me up until 1am:
• It demonstrates why we opted for the rolling average in the first place
• It highlights the problems with the raw data
• It was a useful exercise to undertake that will come in useful in future
Oh, one final thing before I log off:
Courtesy of a brilliant tip-off from @Crick247, here’s a very interesting bit of weekly "seasonality" in reported UK daily deaths:
Every Sunday and Monday, reported deaths are lower than Saturday. Every Tuesday, they rise sharply 
Why?pic.twitter.com/ZQglhKHcsc
My theory: • We know these numbers are allocated to a day based on when they can be reported, not when that person died • We also know they are deaths that occurred before the day of reporting • NHS England only reports a death once family members have been informedpic.twitter.com/KjIeLKwiQL
• Hunch: NHS either make fewer attempts or find it harder to get through to relatives over the weekend (or some other bit of processing slows at wknds) • So despite deaths occurring on Sat & Sun, fewer get processed over the weekend, depressing the reported numbers on Sun & Monpic.twitter.com/rSByEZ8uUM
• And then on Mon, that processing/informing bottleneck is cleared, resulting in a glut of deaths in Tuesday’s report • This happened like clockwork todaypic.twitter.com/ylTMh6hh5K
Almost forgot: dinner tonight (last night?) was an amazing mushroom risotto.
Usually we cook together, but this one was painstakingly prepared while I sat at my laptop growing increasingly exasperated with spline functions.
My girlfriend is the real hero behind the charts 
pic.twitter.com/l9ICqiibaA
Twitter may be over capacity or experiencing a momentary hiccup. Try again or visit Twitter Status for more information.
All charts: 