Results 1 to 2 of 2
2003-10-02, 12:49 #1
- Join Date
- Apr 2001
- Joliette, Quebec, Canada
- Thanked 0 Times in 0 Posts
CALULATION OF POISSON PROBABILTY (97)
To calculate the probability of that 120 back aches in a hospital is really different that the average number of the last 4 years which is a 100 i use = 1-poisson (120-1,100,true) which gives a correct answer of 2,8 % . However I cannot not get a number if the numbers are 10 times that level for = 1-poisson (1200-1,1000,true) . My answer is and I quote #number!. I would like to compute the p for all 10 hospitals. What are the computational limitations on using the poisson function.
2003-10-02, 13:50 #2
- Join Date
- Mar 2002
- Thanked 29 Times in 29 Posts
Re: CALULATION OF POISSON PROBABILTY (97)
The formula for the Poisson distribution involves factorials; the largest number for which Excel can compute the factorial is 170. Excel probably doesn't use the literal formula, but some kind of approximation; still, numbers like 1000 cause an overflow.
For large numbers, the Poisson distribution can be approximated by a normal distribution with the same mean, and with the square root of the mean as standard deviation. So =1-NORMDIST(B3-1;B2;SQRT(B2);TRUE) would be a good approximation of =1-POISSON(B3-1;B2;TRUE). See attached modified spreadsheet.