Dear all,
First I will share some details about my research:
- My dataset consists out of 23 statements.
- I would like to analyze my dataset for two subgroups (Advisers, HCP that do recommend intervention x and Non-advisers, HCP that do not recommend intervention x)
- The aim of my study is to identify differences in how these two groups (slightly) agreed or (slightly) disagreed with the 23 statements.
- Hence, we performed a PCA. This PCA graph of variables and their loadings show us statements advisers/non-advisers agreed upon (most of then (slightly) agreed or (slightly) disagreed, there was 'consensus') and statements they didn't agree upon (there was lots of variation among how they answered on the statement).
- Lots of variance (high loadings) indicating advisors or non-advisers did not reach "consensus" as a group. Hence, that particular statement is probably not related to being an advisor or non-adviser.
- At this moment I am not sure if what I am doing is valid or that I overlook some important points.
- Furthermore, I am wondering if there is a technique on how to compare outcomes of PCA's of two groups within the same dataset?
Kind regards,
Anne
Anne van der Geest
If you do two different PCA and plot them side by side you should see the differences like two different hours in a clock.
But, why don´t you use a decision tree using rpart. Here is the package https://cran.r-project.org/web/packages/rpart/index.html
Just take the example of Titanic data that is attached to the rpart library and change OpnionA - OpinionB instead of Survive-NotSurvive. Play around and may be you´ll get something.
Best!!
- Badges
- Science topic
- Similar topics
- Mathematical Sciences
- Graphs