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
- /
- Burr, Pareto and LogNormal Distributions

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

02-07-2013 11:34 PM

I have a data set of the following figures

Acct Loss

A 250,000

B 325,000

C 450,000

D 125,000

E 680,000

F 110,000

G 997,000

H 500,000

I'd like to produce an output set of total loss by the following distributions at the 50th, 70th, 90th, 95th and 99th percentiles: Burr, Pareto and LogNormal.

In addition, I'd like to produce the K-S statistic, the A-D statistic and KS-Test p-values for the Burr, Pareto and LogNormal distributions (fit statistics for the distributions).

I thought I could use PROC SEVERITY in the following manner:

proc severity data=sample_set;

model loss;

dist burr;

dist pareto;

dist logn;

run;

Unfortunately, the code is not correct.

Any suggestions on how to tackle this challenge would be greatly appreciated. Thanks.

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

02-08-2013 09:39 AM

Your syntax is a bit off. Also, take out the comma from the numbers.

data a;

input Acct $ y;

datalines;

A 250000

B 325000

C 450000

D 125000

E 680000

F 110000

G 997000

H 500000

;

proc severity data=a crit=aicc;

loss y;

dist _predefined_;

run;

Note: you use LOSS to specify the response variable. The keyword _PREDEFINED_ is used to request fits for all the predefined distribution models. But a warning: n=8 is really very small for this type of analysis. It will be difficult to distinguish the models with 8 data points. But the above does "work"..