Talk:Data Mining Algorithms In R/Clustering/Expectation Maximization (EM)

Formula for M-Step
I think some of the formulas are wrong, e.g.

$$ \sum_j(t+1) = \frac{\sum_{k=1}^N\ P(C_j|x_k)\ (x_k - \mu_j(\mathbf{t+1}) )\ (x_k - \mu_j(\mathbf{t+1}) )}{\sum_{k=1}^N\ P(C_j|x_k)} $$

--sigbert (discuss • contribs) 08:07, 13 December 2013 (UTC)

EM isn't the same as clustering, is it?
There seems to be a confusion here in describing the E-M algorithm, a general purpose tool that includes hidden Markov models, Kalman and other filters, K-means clustering, etc. This article is mostly about clustering, but then why not mention K-means explicitly, since this an E-M type algorithm and there's a package in R (base:kmeans) that might be useful to readers.

Also, the R example didn't work for me, choking on the line

Finally, grammar and syntax are not great, making it hard to understand the intent at times.