Find Median of the second column when first column is Date

I have a an Excel file of two columns, First column is date in this format (mm/dd/yy), Second column is observations (number).

I have multiple observations each day and I want to find the Median of these observations for each day separately and save it an Excel file of two columns, first is the day, second is the result (Median).

I start doing this in Excel, but I realized its going to take me forever since I have a large amount of data.

How can I do this in Matlab ?

[Massimo Zanetti]

Post an example of your data and variables, because it is not clear.

[José-Luis]

• Import your data to Matlab xlsread()
• Get the median using accumarray(), or whatever tickles your fancy. The documentation on accumarray is horrid, IMO, but that's a rant for another day.

[Guillaume]

For several versions now, matlab has had readtable which is a lot more powerful than xlsread.

For slightly fewer versions, matlab has had findgroups and splitapply which are easier to use than accumarray (but more or less do the same).

[José-Luis]

Good to know. I should try to keep up.

Answer 1

[Guillaume]

In excel, you would be able to do this easily with a pivot table if only pivot tables supported calculating the median (you can get the sum, mean, standard deviation, and a few others, but not the median!).

In matlab, you'd read your excel file with readtable, then use findgroups and splitapply to Split Data into Groups and Calculate Statistics. Finally writetable back into excel.

Something like:

t = readtable('C:\somewhere\yourfile.xlsx');
[group, dates] = findgroups(t.NameOfDateColumn);
medobvservation = splitapply(@median, t.NameOfObservationColumn, group);
writetable(table(dates, medobvservation), 'C:\somewhere\newfile.xlsx');

[Rawan Aloula]

Saved me a lot of time ! Thank You so much.

Answer 2

[Andrei Bobrov]

if your MATLAB older than R2015b

[n,tt] = xlsread('2008 Data (find daily Median).xlsx');
[y,m,d] = datevec(n(:,1));
[a,b,c] = unique([y,m,d],'rows');
z = accumarray(c,n(:,2),[],@median);
out = [tt;cellstr(datestr(n(b,1),'mm.dd.yyyy')),num2cell(z)];
xlswrite('new_file.xlsx',out);