Here's a well commented example:
m = 10 * rand(3, 24) - 5 % Create sample data (replace with your actual array).
posMap = m > 0 % Get map of where the positive values are.
m2 = m .* posMap; % Initializes m2.
m2(~posMap) = nan % Set negative values to nan so we can ignore them when we take the mean.
% Get mean of each column, meaning you average going down rows.
columnMeans = mean(m2, 1, 'omitnan')
% Get the mean of each row, meaning you average over all the columns in each row.
rowMeans = mean(m2, 2, 'omitnan')
% Get the mean of any positive value, regardless of which row or column it lives in.
overallMean = mean(m2(:), 'omitnan')
