topic Re: Rate calculation using SAS in SAS Programming

Rate calculation using SAS

abdulla — Mon, 03 Feb 2020 04:25:46 GMT

Hi,

I want to calculate the customer arrival rate. A customer arrives in the store in how many minutes. For example, from our data, we may get that a customer arrives at the store in every 7 min.

if we have two columns, one for "minute" and the other for "no of the customer", what would be the SAS code? This is a Poisson distribution. Please help

Minute "Customer of customers"

5 2

10 6

7 3

6 1

4 2

9 5

11 4

2 3

8 5

So, I want to get " In every how many minutes a customer come?". This is the property of Poisson distribution. The above is just an example. The answer could be " a customer comes in every 1.5 min".

Re: Rate calculation using SAS

Kurt_Bremser — Mon, 03 Feb 2020 03:15:20 GMT

Pleaee supply example data in usable form (data step with datalines), and what you expect to get out of it.

Re: Rate calculation using SAS

abdulla — Mon, 03 Feb 2020 04:26:17 GMT

I have edited the question.

Re: Rate calculation using SAS

unison — Mon, 03 Feb 2020 04:57:23 GMT

Look into proc genmod which has Poisson regression capabilities: https://stats.idre.ucla.edu/sas/output/poisson-regression/

Re: Rate calculation using SAS

abdulla — Mon, 03 Feb 2020 05:49:30 GMT

Based on the link which one would be correct

proc genmod data =have;
 model minute = customer / link=log dist=Poisson;
run;

or the following?

proc genmod data =have;
 model customer = minute / link=log dist=Poisson;
run;

Re: Rate calculation using SAS

Kurt_Bremser — Mon, 03 Feb 2020 12:31:00 GMT

Doesn't it mean you have 31 customers arrive over the course of an hour? Then a customer would arrive every 3600/31 seconds on average.

Re: Rate calculation using SAS

Reeza — Mon, 03 Feb 2020 16:08:39 GMT

It depends a bit on what the data means. How is the first column defined? Is that the number of minutes between customers and you have 2 customers at minute 5 and 6 come in 10 minutes later? If so, you're looking for the mean time between events, which is probably just the following:

proc means data=have mean;
weight customers;
var time;
run;