Bin edges of discretize function excluding some bins

Hello.

mtransf is 100x5000 double where I have some values between -1.3 and 1.3 in each column. I add the mean in row 101 of mtransf:

mtransf(101,:) = mean(mtransf, 1);

I want to discretize the data and make categories in row 102 based on the mean values. With if loops in for loop it works fine but solution is not elegant:

% for i=1:5000
%     if (-1.3<=mtransf(101,i)) && (mtransf(101,i)<=(-1.0))
%         mtransf(102,i)=1;
%     elseif (-1.0<mtransf(101,i)) && (mtransf(101,i)<=(-0.8))
%         mtransf(102,i)=2;
%     elseif  (-0.8<mtransf(101,i)) && (mtransf(101,i)<=(-0.6))
%         mtransf(102,i)=3;
%     elseif  (0.8>=mtransf(101,i)) && (mtransf(101,i)>=0.6)
%         mtransf(102,i)=4;
%     elseif (1.0>=mtransf(101,i)) && (mtransf(101,i)>0.8)
%         mtransf(102,i)=5;
%     elseif (1.3>=mtransf(101,i)) && (mtransf(101,i)>1.0)
%         mtransf(102,i)=6;
%     end

I want to do the same task with discretize, but without the bin between -0.6 and 0.6 :

binEgdes = [-1.3 -1.0 -0.8 -0.6 0.6 0.8 1.0 1.3];
catnames = ["1","2","3","4","5","6","7"];
mtransf(102,:)=discretize(mtransf(101,:),binEgdes,"Categorical",catnames);

Thank you.

[Cris LaPierre]

How would you categorize the values between -0.6 and 0.6?

[Filip Simjanoski]

There are non existent.

Answer 1

[Cris LaPierre]

mtransf(102,:) = discretize(mtransf(101,:),binEgdes,"Categorical",catnames);
mtransf(102,mtransf(102,:)==4)=NaN;

[Cris LaPierre]

If you don't even want the space for "4" to appear on the x-axis, then your underlying data can't be double. You need your X values to be categoricals. The challenge with this is that you cannot mix datatypes in a variable. You would need to capture the result of discretize in a new variable of datatype categorical.

Categories do not need to be sequential, so you can remove 4 completely.

dis = discretize(mtransf(101,:),binEgdes,"Categorical",catnames);
dis = removecats(dis,"4");
histogram(dis)

[Filip Simjanoski]

Because there are no values in "4" category this row does nothing. I still get emply bin, like the previous picture. Maybe there is a way to "delete" category?

catnames = ["1","2","3","test","4","5","6"];

[Cris LaPierre]

I explain why in my 2nd post. Use the code from the second example then.

[Filip Simjanoski]

Okay this solved my problem. In the end I can still add the categories in my matrix:

dis = discretize(mtransf(101,:),binEgdes,"Categorical",catnames);

dis = removecats(dis,"4");

mtransf(102,:)=dis;

histogram(mtransf(102,:))

Thank you for the fast reply.

[Cris LaPierre]

That works, but be careful. Your X-axis will be doubles again when you create the histrogram this way, meaning your X-axis is continuous. You will see a bar corresponding to X=4, but corresponds to category "5". Same for X=5 and 6.