World cumulative fit looks good, but daily curve is behind the data which has been chopping up and down lately. It will eventually catch up. Lognormal and neural network plots are poor fits and not with showing for now.
Yesterday was a serious outlier on USA daily deaths data, up 2000 over the previous day. Here are the daily numbers for the last days the last 5 days: [1509 2303 2494 4591 3857]
Chris, all this assumes a homogeneous population. How can you take into account the hotspots that represent different populations/curves? Those will screw up your curve fitting.
ReplyDeleteMike Jervis here
ReplyDelete