0
$\begingroup$

Background: Someone provided me an excel document that they used for classification (claiming that it followed some unsupervised learning algorithm) and asked me to figure out the generalized formula that was utilized for the data to reproduce it. Anyway, we have the following: $$ \frac{(x_i+\max(x_i))^2}{VMR}$$

where each $x_i$ was normalized, $\max(x_i)$ was the maximum value of the $x_i$'s, and the VMR was defined as the variance-to-mean ratio or variance/mean for the variable...

This was then iterated across multiple variables, and the square root of the sum of this iteration was the final formula.

My question is whether any statisticians recognize this as a known statistical process, otherwise it must have just been a formula derived for/from this data.

Thank you for your help!

xx

  • 0
    Some MathJax/LaTeX formatting would make this readable and help you get answers, FYI. – 2017-01-24
  • 0
    Is this possibly what it does $\sum\limits_{i} \mu_i.\frac{(x_i + max(x_i))^2}{\sigma_i^2}$? – 2017-01-24
  • 0
    Hi knk, the argument in the numerator should actually be (x_i - max(x_i)), my mistake on the post, otherwise it looks like it may be equivalent.. is this a known process? Thanks! – 2017-01-24

0 Answers 0