1
$\begingroup$

How would I de normalize the values which where normalized by the min max normalization below ?

enter image description here

$\endgroup$
  • 3
    $\begingroup$ If you know $min_A$ and $max_A$ this is just a simple algebra. If you don't know them nothing could be done. $\endgroup$ – Tim Mar 15 '15 at 10:07
2
$\begingroup$

You are referring to simple feature scaling, lets denote below the scaled $x$ as $x'$. Simple scaling translated the initial range of $x$ into $[0, 1]$ range but it can be simply recoded to any $[a, b]$ range as in your example:

$$ \frac{x - x_{(1)}}{x_{(n)} - x_{(1)}} (b + a) + a = x' $$

The process can be simply reversed by transforming $x'$ back into $[0, 1]$ range and then into new range:

$$ \frac{x' - a}{b - a} (x_{(n)} - x_{(1)}) + x_{(1)} = x $$

Simple arithmetic tells us that if $a + b = c$ then $c - b = a$ and if $gh=d$ then $d/h = g$, so there is nothing fancy in this transformation. However to reverse the scaling you have to know the initial min ($x_{(1)}$) and max ($x_{(n)}$) values because by this operation you can transform range of your data into literally any range, so you cannot deduce the initial scale from $x'$.

$\endgroup$

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.