Sunday, October 4, 2009

Central Limit Theorem


There's this remarkable theorem that goes something like this:

Suppose we record the average daily temperature (measuring the temperature every hour, for 24 hours -- then averaging these 24 readings).
Suppose, too, that we record these daily averages for a few years.
Finally, we plot the frequency with which certain averages occur and get the green distribution shown here:

The red curve is the world famous "Bell Curve".
The fact that the frequency distribution looks like the Bell Curve is interesting, eh?

What's even more interesting is the fact that if we plotted the average height of people we meet each day (or their average weight), we'd still get (very nearly) a Bell Curve.

We might also repeat this ritual for the average of the number of kilowatts used by a city each hour or the average length of fish we catch each day (while fishing each day in Lake Ontario ... and catching lots of fish).

These averages always look like the Bell Curve if there are lots of numbers involved in the average. (That is, we meet lots of people each day or catch lots of fish.)

That's the Central Limit Theorem and (after a jillion years of ignoring it) I decided to explain it ... here.

 

No comments:

Post a Comment