From OpenWetWare
Jump to: navigation, search

Poisson Statistics Summary

When counting random events that rarely occur, the Poisson distribution is theorized to be the parent distribution describing the probability of obtaining x number of counts for a given unit of time. I want to test this theory with real data of counts of random events (I assume these events to be detections of cosmic radiation, but I am not sure). I also want to compare the Poisson distribution to another distribution called the Gaussian distribution for many trials where each trial has a different probability that a count will occur during a unit time.

A large amount of systematic error was present, but I still concluded that the Poisson worked not terribly. There really isn't too much qualitative stuff for me to report. But I can report that when the average counts per unit time was small, 0.077(5), the (Poisson Goodness of Fit)/(Gaussian Goodness of Fit) was about 5. Measuring (Goodness of Fit) was difficult for me, and I could not think of a way to report its uncertainty.

For a much better explanation of this lab and cool graphs, see the following link:

link to my Poisson statistics notebook