Measuring performance by MSE or RMSE in classification/clustering tasks?

Question

I employ K-means and MLP algorithms for two simple clustering and classification tasks. I searched many in the literature and I found that some researchers applied MSE and other RMSE for comparing methods and their results.

Is there any logical and theoretical difference between MSE and RMSE in clustering/classification performance measurements?

For instance, if our dataset is normalized between [0...1] or is not normalized, which one is appropriate? Does MSE/RMSE depend on the normalization of features? or any scale?

Answer 1

RMSE is the square root of the MSE.

Since the square root is a monotone function, you'll get the same ranking. Just the number has a different interpretation. RMSE can be more meaningful when you understand the data.

Don't use it for clustering. Use it for classification and regression only.