topic Re: Suggest a code/method for logistic regression in Statistical Procedures

Suggest a code/method for logistic regression

Sathish_jammy — Wed, 09 Sep 2020 06:00:17 GMT

Dear All,

I have a data set of patients were consuming drugs on given date. The table categorized by cluster variable. I need to find which drug influence/significance more in each cluster type.

for example : (Cluster_Type - Drugs)

Cluster1 - Bigu, Sulp, DPP4, Insulin, Sulp,Bigu,

Cluster2 - Bigu, Sulp, DPP4, Insulin, Bigu, Sulp

Cluster3 - Bigu, Sulp, DPP4, Insulin, Insulin, DPP4

I attached the csv dataset for your reference.

Kindly suggest a code to solve the problem.

Thanks in advance!

Re: Suggest a code/method for logistic regression

PaigeMiller — Wed, 09 Sep 2020 11:18:14 GMT

I don't download attachments any more. Please provide a portion of the data as a SAS data step (via these instructions: https://blogs.sas.com/content/sastraining/2016/03/11/jedi-sas-tricks-data-to-data-step-macro/)

You have not mentioned a response variable. Please describe that part of the problem.

Re: Suggest a code/method for logistic regression

ballardw — Wed, 09 Sep 2020 16:37:04 GMT

@Sathish_jammy wrote:

Dear All,

I have a data set of patients were consuming drugs on given date. The table categorized by cluster variable. I need to find which drug influence/significance more in each cluster type.

for example : (Cluster_Type - Drugs)

Cluster1 - Bigu, Sulp, DPP4, Insulin, Sulp,Bigu,

Cluster2 - Bigu, Sulp, DPP4, Insulin, Bigu, Sulp

Cluster3 - Bigu, Sulp, DPP4, Insulin, Insulin, DPP4

I attached the csv dataset for your reference.

Kindly suggest a code to solve the problem.

Thanks in advance!

I would say your "cluster" definition needs some clarification as well. Is that list supposed to be the order of administration or something? Cluster typically refers to group membership in statistics and having an item repeated doesn't make sense OR having something repeated even though the items are the same making a different "cluster".

Not even mentioning the bit that none of the values except Insulin that you mention in your "cluster" definition even appear in the data. I am not going to guess about what should be done.

What role do the Group variables play in this?