If you have 2 addresses that are making transactions during similar times and always use the same fee-rates then it is very likely that they are owned by the same person.
If you're using the same address multiple times, that's a bigger privacy concern than the fees you're paying. Given the number of transactions out there, I'm not concerned about linking 2 random transactions. Mempool.space usually shows (many) hundreds of transactions with similar fees per block.
I don't think you fully understood me or I failed to express myself correctly. I used the address reuse example as in that case the model will give a stronger probability, but it also works with new addresses. If you reuse 2 addresses that are in two completely separate wallets there is a general risk that comes from address reuse for both clusters. However, linking these two wallet clusters is a completely separate risk which has nothing to do with the previous one. Further, nowhere did I imply that fee-rates are a bigger concern than address reuse. Lastly, the model does not work alone on fees. Fees are part of the model, and it can work quite well. You should definitely always pay different fees using different fee-rates between transactions. It effectively costs you nothing time wise, but provides a small padding for your anonymity.
There are probably continuously running statistical models doing analysis with different weighs, and they will include fee-rates to some degree.
I bet they'll try, but it's going to be one big guessing game with at best a low probability of 2 addresses being owned by the same person, and many false positives.
No, the models work pretty well otherwise nobody would need to use Monero.
As I said, there is a misunderstanding. Anything you do differently will help you a bit against such models. I'll leave it at that to avoid going more off-topic.
