R中的因素分析，误差

Question

I have typed the following command in R

factanal(TD, factor = 10, rotation = "varimax", scores = "regression")

where TD is my data frame containing 41 variables

I am getting the following error

Error in factanal(TD, factor = 10, rotation = "varimax", scores = "regression") : unable to optimize from this starting value

Can someone help me with this error.

Answer 1

I have had the same problem, I found that adding the control argument "lower", which specifies the lower bound for uniqueness (default = 0.005), helps!

For my problem a value of 0.00000001 gave results. Afterwards, it turned out that I had some variables with very low values of uniqueness. So it might be that you have created 4 dummies from a factor with 4 levels, thereby creating redundancy.

So my tips: 1. Play with the lower argument 2. once it works, check which variables have a very low uniqueness, take them out of the dataframe & run the factor analysis again

Answer 2

There is a "start" option in factanal like this:

factanal(TD,factor=10,rotation="varimax",scores="regression",start=rep(0,41))

But without your data/detailed output can't tell what exactly the problem is.