How to calculate the frequency of a binary event (0 or 1) on a weekday basis over the last 10 years?

I have a single column Logical matrix with 0 and 1's and a single column weekday char matrix with weekday names. I want to calculate how may times I get a 1 for each weekday for the entire period of both matrices (both are about 10 years). Then the same for 0. Then Id like to visualise it. Any ideas?

Answer 1

[Azzi Abdelmalek]

n=20
a=randi([0 1],n,1)
b={'monday';'tuesday';'Wednesday';'Thursday';'friday'}
w=b(randi(5,n,1))
v=[num2cell(a)  w]
[ii1,jj1,kk]=unique(v(:,2))
out1=[ii num2cell(accumarray(kk,a))]
out0=[ii num2cell(accumarray(kk,~a))]

Answer 2

[Andrei Bobrov]

t = datenum([2015 2 14;2015 8 15]);
[N,S] = weekday((t(1):t(2))');
O = rand(numel(N),1) > .5;

% ones

out1 = accumarray(N,O);

% zeros
out0 = accumarray(N,~O);

[~,b] = unique(N);

out = [cellstr(S(b,:)), num2cell([out1,out0])]

Answer 3

[Walter Roberson]

Letting Lvec be the logical column vector, and letting Wvec be the character column vector, then

dayorder = 'MTWJFSD';   %the day characters in the order you want to present

%need to translate day characters into bin numbers for counting. But
%potentially some day characters do not appear at all, and potentially
%some wrong characters appear. We could first check for validity and then
%bin against the known characters, but the validity check can give us bin
%numbers as a side-effect, so might as well use them. The bin numbers
%will not match the order of the days, but we can find the map between the
%two at the same time we are checking if any entries were invalid

[unique_daychar, ia, binnum] = unique(Wvec);
[tf, ordermap] = ismember(unique_daychar, dayorder);
unknown_days = unique_daychar(~tf);
if ~isempty(unknown_days)
  error(sprintf('Unknown day codes in input: %s', unknown_days(:).'));
end

%so know we have bin numbers with the bins in an arbitrary order (actually
%in the order the characters sort into), and we have a mapping between that
%arbitrary order and our desired presentation order. We can do stats in
%either order. It's just easier to present the output if we do the stats
%in presentation order. That's why we map the bin numbers that we send into
%the counting routine, so that the counts come out in presentation order

stats = accumarray([1+Lvec, ordermap(binnum)], 1, [2, length(dayorder)]);

%the rest is just visualization, make it as pretty as you like
fprintf('Day order: %s\n', dayorder);
fprintf('Counts for 0:\n');
fprintf('%5d', stats(1,:));
fprintf('\nCounts for 1:\n');
fprintf('%5d', stats(2,:));
fprintf('\n');