create empty row inside a vector

Hi. I need to analyze the rows of the vector 'value_GS_test' and transform only the numbers 254 and 255 (if any) into '[]' while keeping the same number of rows as in the original vector. Here is an example:

Right now I am using this code, but it only deletes one row (at 255) for me. How can I get the desired result?

value_GS = importdata("value_GS_test.mat");  % 6 elements

for rows_value_GS = 1 %:height(value_GS)
    
    analysis_number = value_GS(rows_value_GS);
    
    if analysis_number == 255 
    
        value_GS(rows_value_GS) = [];

    elseif analysis_number == 254 
    
        value_GS(rows_value_GS) = [];
    
    else
        value_GS(rows_value_GS) = value_GS(rows_value_GS);
            
    end

end

[Chunru]

How about using nan instead of []?

[Stephen23]

"How can I get the desired result?"

Numeric arrays cannot have empty elements or "holes" in them: every element is one value.

You could use a marker value, e.g. NaN, or store a separate logical mask.

[Walter Roberson]

Note however that uint8 cannot represent NaN or inf

[Simon Chan]

What is the final purpose of making a bracket to those positions? For display only? Or you will write it as blank on a text file?

[Alberto Acri]

For display

Answer 1

[Chunru]

load(websave("value_GS_test.mat", "https://www.mathworks.com/matlabcentral/answers/uploaded_files/1436593/value_GS_test.mat"));

value_GS = double(value_GS)

value_GS = 6×1

255 255 255 68 70 72

value_GS(value_GS>=254) = nan;
value_GS

value_GS = 6×1

NaN NaN NaN 68 70 72

[Alberto Acri]

Thank you for your reply. I prefer [] instead of '0' or 'NaN'.

[Chunru]

If you prefer '[]', you may have to use string, which may not be efficient.

[Stephen23]

A string will most likely be complex and inefficient. Perhaps a mask would work... but it depends entirely on what they are doing with this data. So first the OP needs to answer this very important question:

https://www.mathworks.com/matlabcentral/answers/1997313-create-empty-row-inside-a-vector#comment_2819578

[Alberto Acri]

Thank you for the clarification @Stephen23

[Alberto Acri]

Hi @Chunru. Is it normal that with your code I am displaying '0's instead of 'NaN'?

[Dyuman Joshi]

Yes, it is.

Refer to @Walter Roberson's comment above - "Note however that uint8 cannot represent NaN or inf".

When you try to assign NaN (or Inf for that matter) to a uint8 (or any integer data type) array, it will display 0 instead of NaN

%Signed integer data type example - 
y = int16(1:4)

y = 1×4

1 2 3 4

%Assinging 2nd element of the vector to be NaN
y(2) = NaN 

y = 1×4

1 0 3 4

%Unsigned integer data type example -
%Initializing NaN with an unsigned integer leads to 0 as well
z = uint8([1 23 NaN 19 5])

z = 1×5

1 23 0 19 5

[Alberto Acri]

Okay. Thanks for the explanation @Dyuman Joshi. I am unfamiliar with Matlab.