Estimating a distribution question

udt1

New Member
#1
Hi,

I have information on the first, third, and fourth quartile of a distribution, and a sample that consists of about 20% of a total population. I am looking for a way to infer the likely distribution of the entire population. Can anyone point me in the right direction?

Thank you,

udt1
 

mp83

TS Contributor
#2
First of all you should plot your data.Tehn see if a known pdf is likely to fit them.If then you should use the quantile matching method. MAthch the theoretical quantiles to the sample's quantiles.

If you cannot find a candidate distribution then things are tougher.
 

udt1

New Member
#3
Hi,

Thank you for your response. The data more or less fit a normal distribution, truncated on one side due to a limit in the maximum value.

I imagine that the quantile matching method must involve shifting the data and/or filling in hypothetical data points to fit the distribution? Is there software available to do this?

Thanks again.
 

mp83

TS Contributor
#4
I think R has function(s) on quantiles estimation on its base installation (?quantile)

But for a truncuated distribution I think you'll have an issue coz that's not astandard procedure.Anyway try it and then let the topic know what happened