0
$\begingroup$

I'm currently working on a project where I have a dataset which consists of a number of blood samples and the quantity of different biological compounds within each sample.

The samples are split into three groups - severe disease, mild disease and controls.

My aim is to try and find which of these compounds have a significant relationship with the severe disease and which have a significant relationship with the mild disease in comparison to the controls. This means if I was going to write up my results I would need to be able to see the individual names for each compound (currently column names in the dataframe).

I'm hoping to use logistic regression but I'm not sure how to go about it in such a way where I can pull names for these specific compounds.

I also am aware that I should be reducing my dataset in some way, but surely if I complete a regression analysis with principle components I won't be able to determine which individual compounds are significant?

$\endgroup$
1
  • $\begingroup$ If you use PCA getting back to the features and their contributions is much harder. If inferring is very important to your problem, compared to prediction, it might be better to avoid PCA. I rarely use PCA in the model so I can infer from the features. $\endgroup$
    – Craig
    Aug 4, 2022 at 15:34

1 Answer 1

0
$\begingroup$

Using some of R packages, you may see the p-value by each feature. For numeric features, it will present for the whole column. However, to categorical features, it will present for each factor. See the model fitting on this link. https://www.r-bloggers.com/2015/09/how-to-perform-a-logistic-regression-in-r/

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge that you have read and understand our privacy policy and code of conduct.

Not the answer you're looking for? Browse other questions tagged or ask your own question.