To find the distribution pattern

[shruthi]

I have 103 years of daily rainfall data. I need to know whic
probability distribution it follows. Is it possible to find th
distribution pattern followed by rainfall using MS excel

 

[Jerry W. Lewis]

I know of no program where you simply plug in the data and it spits out a
consonant distribution. In many programs (including Excel) you can assess
the goodness of fit of the data to a particular hypothesized distribution.
Stat packages would have more options for this than does Excel, but two Excel
based approaches that immediately come to mind are a Chi-Square goodness of
fit test or a probability plot. Mike Middleton has directions for producing
a Normal probability plot
http://www.usfca.edu/~middleton/data.htm
that could be adapted to more appropriate distributions. I would be
surprised if there is not a standard probability model for rainfall, but I
have never worked with rainfall data. Barring that, since negative rainfall
is impossible, I would begin by investigating the lognormal or gamma
distributions.

Jerry

 

[Mike Middleton]

shruthi -

I strongly recommend that you summarize time sequence data using a
probability distribution _only_ if you have checked to see that the time
series is stationary, i.e., same mean and variance over the entire time
period with no patterns.

I suspect that there is strong seasonality (repetitive pattern) in the daily
observations, so you will lose considerable information if treat all
observations as independent and identically distributed. I suggest using
methods of time series analysis.

- Mike
http://www.mikemiddleton.com