turn on suggestions

Auto-suggest helps you quickly narrow down your search results by suggesting possible matches as you type.

Showing results for

Find a Community

- Home
- /
- Analytics
- /
- Stat Procs
- /
- Fitting censored data to Possion distribution

Topic Options

- Subscribe to RSS Feed
- Mark Topic as New
- Mark Topic as Read
- Float this Topic for Current User
- Bookmark
- Subscribe
- Printer Friendly Page

- Mark as New
- Bookmark
- Subscribe
- Subscribe to RSS Feed
- Highlight
- Email to a Friend
- Report Inappropriate Content

02-12-2016 11:51 PM - edited 02-12-2016 11:52 PM

Hello there,

I am planning to fit some data with interval CENSORING in SAS. I did that with severity and it was amazing. Unfortunatly, I could not find Poisson distribution in the supported distributions by the Severity Proc. My data are collected from cars and it is hard to judge if it is continious or discrete. Hence, I have to try famous distributions and the only one which is left is poisson. I appreciate that if you could kindly help me with that. I need QQ, Probability plot, and AIC as outout. Any help is highly appreciated.

Kind regards,

Mohsen

My current code for Severity:

I did left censoring for values less than 3 and right censoring for values greater than 500 (kind of interval censoring)

data df;

infile "/home/Username/PathAll_TWOMONTH _BothDirection715_2.csv"

LRECL=10000000 DLM=',' firstobs=2;

input

timenumn count right left

;

run;

data t1(drop=count left right rename=(timenumn=t1)) ;

set df;

do i = 1 to count;

output;

end;

run;

data t2(drop=count left right rename=(timenumn=t2));

set df;

do i = 1 to count;

output;

end;

run;

data count(drop=count left right rename=(timenumn=count)) ;

set df;

do i = 1 to count;

output;

end;

run;

data t1;

set t1;

/*If t1 < 3 then t1 = 1;

if t1>1000 then t1=1000;*/

if (t1<3 ) then t1=1;

if(t1>2) then t1=t1;

if(t1>500) then t1=500;

run;

data t2;

set t2;

/*If t2 < 3 then t2 = 5;

if t2 >1000 then t2=t2;*/

if(t2<3) then t2=3;

/*if(t2<501) then t2=t2;

if (t2>500 ) then t2=500;*/

run;

;

/*

data count;

set count;

if count <100005 then count=1

run;

;

*/

data final;

merge t1 t2 count;

run;

proc severity data=final print=all plots(histogram kernel)=all

criterion=ad;

loss count/ lc=t2 rc=t1; /* <=alan doroste, rc=low better */

dist _predef_;

run;