0
$\begingroup$

If I have various values, $x,\, y,\, z,\, t$. And I want to give each of these values a weight such as $x(30),\,y(10),\,z(40),\,t(20)$. This weight relates to that values part off $100\%$, so if each value was the most it could be and we added all values it would be 100. How can I make this type of equation ? for example in the event that $x$ was $50\% x = 30$, y was $100\%y = 10$, and $z$ and $t$ are $50\%$ I could get a score of $80$.

So basically I want the answer to be a value less or equal to $100$ were the values that are added to get the score are weighted and take up different portions of the overall score.

$\endgroup$
0
$\begingroup$

If you split a number $s$ into 4 parts with relative proportions $x:y:z:t$ you get $\frac{xs}{x+y+z+t}$, $\frac{ys}{x+y+z+t}$, $\frac{zs}{x+y+z+t}$, and $\frac{ts}{x+y+z+t}$.

Let's check this is correct:

It is obvious that these numbers are simply $x,y,z,t$ multiplied by $\frac{s}{x+y+z+t}$, so they have the right proportions. Their sum is $\frac{xs+ys+zs+ts}{x+y+z+t} = \frac{(x+y+z+t)s}{x+y+z+t} = s$ as required.

$\endgroup$
3
  • $\begingroup$ I think this is a bit different to what I need ill try to explain better. I have four independent values, I want to add these values and get a final value from 0 - 100. But I want the different values to have different weightings. So I want for instance the first value to make up 30% of the final value, the second to make up 20% and so on. $\endgroup$ – dmnte Nov 29 '17 at 4:04
  • $\begingroup$ @dmnte Are these four independent values also in the range 0-100? If so, just take their weighted average: $(w_1x+w_2y+w_3z+w_4t)/(w_1+w_2+w_3+w_4)$. If not, you'll have to scale those independent values first to make them fall in that 0-100 range before taking the weighted average.. $\endgroup$ – Jaap Scherphuis Nov 29 '17 at 7:47
  • $\begingroup$ After a little searching ive found the softmax function which seems like I could use. It is not weighted but I assume I could add this. Does anyone know anyreasons why this wouldnt be good ? $\endgroup$ – dmnte Nov 30 '17 at 1:23

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.