How do we do operations directly on table elements in matlab?

I am thinking of doing min max normalization of my data where if x is vector the procedure is as follows x = (x- min(x))/(max(x)-min(x)). However I want to do this operation directly on the table elements which have 150 rows and 42 features and I want to normalize each of these columns. Is there any one line procedure to do it without using for loop or something like that ? I know we could use curly and dot bracket indexing but the result is not gonna be a table though.

Answer 1

[Walter Roberson]

normalized_data = normalize(data, 'range'); %works on a table, normalizes to [0 1]

Answer 2

[bharath pro]

If you can use z-score instead of min max, then you can directly do

data=normalize(data);

If you cant do that, then this works in one line.

data = (data - min(data, [], 1)) ./ max(data - min(data, [], 1), [], 1)

[D_coder]

no this doesn't work. The second line i mean,

It says

Invalid data type. First argument must be numeric or logical , and

This makes sense because min(data, [], 1) doesnot give you a table.

the first line is not useful as I want to do min max normalization

[bharath pro]

I missed the table part. If your table is homogenous you can convert to array do the computation and convert it back.

data = array2table((table2array(data) - min(table2array(data), [], 1)) ./ max(table2array(data) - min(table2array(data), [], 1), [], 1))

This is a workaround and not a direct answer of normalizing table entries.

[D_coder]

again I wanna preserve the table columns names I doubt this would do it

[Walter Roberson]

data = array2table((table2array(data) - min(table2array(data), [], 1)) ./ max(table2array(data) - min(table2array(data), [], 1), [], 1), 'VariableNames', data.Properties.VariableNames);

However, the solution I posted using normalize() works directly on tables without that bother.