How do I Convert Cells containing Binary Values into Cells of Decimal Values and Decimal Values into Strings?

Hi Guys,

Let's say I've a Matrix A with 5 Rows and 1000 Columns, consisting mostly of 0s and 1s. Maybe every ~50th or ~100th Column or so contains a Number

other than 0 or 1. Their distribution seems to be totally random. They're infrequent, but they're there.

Here's the thing:

I'd like to convert those Columns of 0s and 1s from their 5 digit binary values into a decimal value. So,

A = mat2cell(A,size(A,1),ones(1,size(A,2))); %Convert A to Cell Array with each Column its own Cell
B = cellfun(bi2de(A')) %Convert the Content of each Cell from Binary to Decimal

would, I guess, do the trick, if it weren't for the monkey wrench thrown in there of those occasional Non-Binary Values.

Is there a function to tell Matlab to, if a cell contains only 0s and 1s, convert its content into Decimal,

but if it contains other integers, simply output "Unknown", yielding a Cell array with the same number of cells as the Input Array.

Kind regards,

Tim

Answer 1

[Image Analyst]

Try this:

% Create sample data.
A = zeros(5, 1000);
A(randperm(numel(A), 200)) = nan;
A(randperm(numel(A), 800)) = 1;
[rows, columns] = size(A);

% Find out which columns have nan's so we can skip them.
columnsWithNan = any(isnan(A), 1);
numbers = zeros(1, columns); % Preallocate output
% Now process getting decimal number for each column
for col = 1 : columns
	if columnsWithNan(col)
		continue;
	end
	thisCol = sprintf('%d%d%d%d%d', A(:,col));
	numbers(col) = bin2dec(thisCol);
end

[William Collants]

works like a charm, thank you kindly 💚

I Put a tini tiny spin on it so the Output Vector has NaNs instead of Zeros where a Column was skipped.

columnsWithNan = any(isnan(A), 1);
numbers = 9999*ones(1, columns); % Here

for col = 1 : columns
	if columnsWithNan(col)
		continue;
	end
	thisCol = sprintf('%d%d%d%d%d', A(:,col));
	numbers(col) = bin2dec(thisCol);
        numbers (numbers == 9999) = NaN; %and Here
end

[Image Analyst]

If that's what you wanted, you would just simply instantiate numbers to nan's rather than the less direct way (more kludgy) of instantiating them to 9999 and then checking if all elements encountered prior to the current element are still 9999 and setting them to nan. And of course removing comments is almost never a good idea because they explain the code so that later when you look at it you'll understand what you did.

% Find out which columns have nan's so we can skip them.
columnsWithNan = any(isnan(A), 1);
numbers = nan(1, columns); % Preallocate output
% Now process getting decimal number for each column
for col = 1 : columns
	if columnsWithNan(col)
		continue; % Skip this element if it's a nan.
	end
	thisCol = sprintf('%d%d%d%d%d', A(:,col)); % Get binary string.
	numbers(col) = bin2dec(thisCol); % Convert string to a number.
end

[William Collants]

oh totally that kinda thing keeps cropping up in my code all over the place, doing in 2, 3, 5, 89 steps what can be done in 1. Hopefully that'll all get weeded out as I become more adept at Matlab over the coming years.

Thank you for the correction. Good protection against the formations of bad habits ☝️

Kind Regards,

T