0
$\begingroup$

I have to make a linear transformation of a mean to "denormalize" it after a prediction comes out of a model. If the denormalization is just a linear transformation of the mean, does the standard deviation need to undergo any transformation as well?

The function used to normalize, (and reversely denormalize) is written below...

$$ normalized = \frac{x - min}{max - min} * (max_{new} - min_{new}) + min_{new} $$

$\endgroup$
1
$\begingroup$

The standard deviation does not change when you add a constant, but it is affected when you multiply by a factor (gets multiplied by the absolute value of the factor). In your case $$ \sigma (normalized)=\frac{max_{new}-min_{new}}{max-min}\sigma(x). $$

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.