Search code examples
matlabartificial-intelligencenormalize

How to normalize a table to range [0;1]

I have a table in matlab , and I want to normalize data between 0 and 1 this is a row of my table:

6 148 72 35 0 33.6000000000000 0.627000000000000 50

What is the best way to normalize data, and which one is better, normalize by each coloumn or normalize all cell that influence together ?

EDIT on 5-24-2016 i just needed to normalize my data to get specific range of value

Solution

  • Normalization could either mean (a) data obtained on different scales are aligned/adjusted in order to enable comparison, or (b) try to align the statistical distribution of data to Normal distribution.

    In this dataset, if you have to normalize, then it is column-wise. Mean and variance seem to be important features here though. To normalize column-wise:

    m = max(D); % D is your dataset matrix
normD = zeros(size(D)); % Here are your normalized values
for i = 1:length(D)
    normD(:,i) = D(:,i)./m(i)
end

    A faster way to do this is without loops: D./repmat(max(D),size(D,1),1)