CODEWITHSUNDEEP
pythonperlmathgraphstatistics
Amiya Behera
Amiya Behera

Reputation: 2270

re-scaling the values between given maximum and minimum, where max is positive and min is negative

I want the re-scale an array of values (both positive and negative floating point values), so that the values lies between say -5 to +5 or say -3 to +2. How can I scale it down. Is there any specific formula for it. I tried searching, but did not found anything that can deal with negative min value.

Thanks in advance...

Upvotes: 3

Views: 11842

Answers (3)

TheMerovingian
TheMerovingian

Reputation: 648

The easiest way to scale your data down is to determine the maximum value of your data (positive or negative) and use that to scale all the other data accordingly.

For example, if you had a maximum of 320 in your data

 320 / 320 * 5  => 5
 160 / 320 * 5  => 2.5
-160 / 320 * 5  => -2.5

So as a general equation

x / abs(maximum) * scale

Upvotes: 2

Pep_8_Guardiola
Pep_8_Guardiola

Reputation: 5252

You could use a linear interpolation/extrapolation formula to get the results you want.

Code:

def rescale(val, in_min, in_max, out_min, out_max):
    return out_min + (val - in_min) * ((out_max - out_min) / (in_max - in_min))

print(rescale(4, 4, 20, 0, 100))  # => 0.0
print(rescale(12, 4, 20, 0, 100)) # => 50.0
print(rescale(20, 4, 20, 0, 100)) # => 100.0
print(rescale(0, 4, 20, 0, 100))  # => -25.0

Note that this code doesn't check your values against the min and max bounds supplied, hence it will extrapolate as in the last example. Modify it as you need.

Upvotes: 3

duffymo
duffymo

Reputation: 308868

Try something like this to map values onto a (0, 1) scale:

y = (x - xmin) / (xmax - xmin)

You can also use finite element shape functions: 

y = h1*x1 + h2*x2

where

h1 = (z - 1)/2.0 for -1 <= z <= +1
h2 = (z + 1)/2.0 for -1 <= z <= +1

This maps your values onto a (-1, 1) scale.

Upvotes: 3

Related Questions