Normalization of multiple variables in MATLAB

I am using 5 inputs and one target variable for my task, I need to know that either we normalize each input variable for its own max and min values or global max and min values.

To make it more clear

x1 = [1;2;3;4;6] ; x2 = [0.5;0.9;8;4;6] ;  y= [32;12;21;43;52];
[x1, x2, y]

ans = 5×3

1.0000 0.5000 32.0000 2.0000 0.9000 12.0000 3.0000 8.0000 21.0000 4.0000 4.0000 43.0000 6.0000 6.0000 52.0000

norm_x1 = (x1 - min(x1)) / (max(x1) - min(x1))

norm_x1 = 5×1

0 0.2000 0.4000 0.6000 1.0000

Is this correct or should I be using min and max values out of all three columns.

[Rik]

Your question is not a Matlab question. It is a question about your application. What exactly is required in your application? What makes sense?

It is easy to implement a global normalization, but whether you should be doing that, we cannot answer.

[Benjamin]

Hello,

Your message is a bite unclear, what do you wanna normalize ?

Each column of [x1,x2,y] ? Or the whole matrix [x1,x2,y] ?

And by what ? By the norm of one of the vector x1,x2 ? Or by the global max and min of [x1,x2,y] ?

Regards,