PDA

View Full Version : problem with fitting exponential distribution


vikrant
01-22-2010, 01:08 AM
Hi All,
This is my first post in this forum.
I have two questions :-
1) I have a data of about 11937 data points. When i plot a histogram for this data almost all of the data is in the first bin. So clearly there are outliers in my dataset. How to remove outliers ? Is there any method other than boxplot for detection of outliers?
2) From the histogram i have removed some points so that my first bin is spread all around. Visually I the data seems to follow a exponential distribution. I have fitted a exponential distribution (rate = 0.024).Visually it seems to be a good fit. but when I use chisquare test for goodness of fit it rejected it. Also the kolmogrov Smirnov test rejected it
I have attached the graph for histogram and fitted distribution. Please have a look at this and suggest which distribution will be suitable ?

omega
02-12-2010, 01:25 AM
The distribution is extremely skewed to the right. You can remove the outliers by excluding values greater than a certain number.

You'll need to use maximum-likelihood fitting to find the distribution and parameter.