Wednesday, October 15, 2014

Exponential? (part Deux)

Above is a homebrewed chart of the number of Ebola cases as reported by the WHO.  I ran the numbers out to the first of the year.  If the numbers start going up faster than these projections, then we may well be in a true exponential growth scenario.

Note to readers:  The orange line is the confirmed, suspected, and probable cases.  The blue line is lab-confirmed cases

There is a lot of issues with under-reporting, the ability of labs in a hot zone to deal with samples, governments and peoples trying to skew the data in their favor, bureaucratic incompetence, and other issues to many to number.

But these are the numbers available to me at the current time.

I may have made errors in data entry, this was just for my own edification.

Right now, the numbers show a tight linearity.  R-values above 0.98 are pretty strong.

This is a plain vanilla exponential curve, note the long flat run-up that looks pretty linear
So, a final caveat.  Exponential functions are tricky little bits.  They can go a long way looking like innocent little linear functions.  No need for panic right yet, but with something like this, it is definitely a good idea to keep a weather eye.

What you need to keep an eye on is the series value vs the predicted value in the system.  Right now, the series value (blue dot) is lower than the predicted value (orange dot) if the series values start breaking consistently above the predicted value and start getting bigger as they go, then Katie bar the door.

Add On at 18:38 10/15/14

Here is the exponential curve fit from Wolfram Alpha

I don't find this that convincing.

When I keep plotting in a standard excel chart and push out linear, I get something that pleases my eye more.

No comments: