Thursday, September 3, 2009

Binomial Statistics


You may think of this as you wish, but I once read a great book that said, if you suspect the process to be binomial (like a plinko game!), then you can expect excursions of 2-3 times, and should not over-steer them. This applies to sales results, profits, quality, etc. In other words, it might just be the nature of the distribution, and nothing you can do.


Harbles said...

I don't think 2k years is a long enough time frame to see any periodicity in Arctic (global) climate.

Sorry if I have posted this before but this guy thinks climatic change may be driven (in part) by cosmic rays. He has an experiment that will try to quantify the effects of high energy particle passage on cloud formation.

Harold Asmis said...

There are many, many variables, that is why it is a plinko game. You could imagine a graph of M'al overall commuting effort, normalized. As bridges collapse, construction increases, there are ups and downs, according to many variables. Then, all of a sudden, everything aligns, the bridges collapse more, more construction, more trucks blowing up, and there is an expected binomial spike. You should not then oversteer with a massive effort, because that would oversink the commute effort, and cause new bedroom communities, greenspace destroyed, etc.