search

jinseob2kim / jstable

Regression Tables from 'GLM', 'GEE', 'GLMM', 'Cox' and 'survey' Results.
https://jinseob2kim.github.io/jstable/
Apache License 2.0
23 stars 13 forks source link

I'm wondering if the function(TableSubgroupMultiGLM) could support family = 'binomial( link = logit)". #19

Closed chuckbyron closed 8 months ago

chuckbyron commented 9 months ago

Dear author.thanks for your excellent R package, but i have one question:

I am analyzing cohort data, and I wish to analysis RR instead of OR value in subgroup analysis.

TableSubgroupMultiGLM( formula, var_subgroups = NULL, var_cov = NULL, data, family = "binomial", decimal.estimate = 2, decimal.percent = 1, decimal.pvalue = 3, line = F )

I'm wondering if the function could support family = 'binomial( link = logit)".

thank u very much!

jinseob2kim commented 9 months ago

family = "binomial" is equal to family = 'binomial( link = logit)"

chuckbyron commented 9 months ago

family = "binomial" is equal to family = 'binomial( link = logit)"

Dear author.thanks a lot for your reply.

I hope to calculate the RR value, instead of OR value. When I use TableSubgroupMultiGLM() function, I got the OR value.
And the readme file in your GitHub also shows the OR value. Could you kindly support the methods to calculate RR value.

thans you very much, and best wishes!

jinseob2kim commented 9 months ago

Currently, jsmodule can't support RR(poisson or quasipoisson). I will update RR until next CRAN release. Thanks for your feedback

jinseob2kim commented 9 months ago

I update RR to TableSubgroupMultiGLM

chuckbyron commented 9 months ago

Thanks for your reply, and the updated package could calculate the RR value in poisson distribution regression. But I still have one request, could you try to support log-binomial regression, sorry to bother you, I meant family = 'binomial( link = log)".

Best wishes!

jinseob2kim commented 9 months ago

I've never used binomial(link=log) before. Can you share some example data and code?

chuckbyron commented 9 months ago

Thanks a lot, here are two websites use binomial(link=log) to calculate RR for log-binomial regression model.

https://stackoverflow.com/questions/66593141/log-binomial-regression-for-binary-outcome-with-multiple-category-predictors-and

https://stats.oarc.ucla.edu/stata/faq/how-can-i-estimate-relative-risk-using-glm-for-common-outcomes-in-cohort-studies/

jinseob2kim commented 9 months ago

Thanks for the link. Sorry, I've never used the "log" option, so it's hard to apply.