Divide timeseries to monthly ones
Show older comments
For a specific year I have created an hourly timetable using retime.
How can I divide into monthly ?
Accepted Answer
A for loop is the easiest way to do this —
LD = load(websave('dataset','https://www.mathworks.com/matlabcentral/answers/uploaded_files/1255052/dataset.mat'));
T = LD.TT1;
VarPerHour = retime(T, 'hourly', 'sum')
for k = 1:12
MMidx = month(VarPerHour.date_time) == k;
VarPerHourMonth{k,:} = VarPerHour(MMidx,:);
end
VarPerHourMonth
VarPerHourMonth{1}(1:5,:)
VarPerHourMonth{12}(1:5,:)
This uses an existing timetable. It should work with the one you are currently using as well.
.
10 Comments
This was an adventure!
This looks a bit more complicated than it actually is.
I can’t get them exactly in the format you want, however this is reasonably close.
In any event, it’s the best I can do —
LD = load(websave('dataset','https://www.mathworks.com/matlabcentral/answers/uploaded_files/1255052/dataset.mat'));
T = LD.TT1;
VarPerHour = retime(T, 'hourly', 'sum')
for k = 1:12
MMidx = month(VarPerHour.date_time) == k;
VarPerHourMonth{k,:} = VarPerHour(MMidx,:);
end
for k = 1:12
TTTemp = VarPerHourMonth{k}; % Create Temporary 'timetable'
Hours = hour(TTTemp.date_time); % Create 'Hours' Variable
[y,m,d] = ymd(TTTemp.date_time); % Begin To Create 'Date' Variable
Date = datetime(y,m,d); % Finish Creating 'Date' Variable
TTTemp = addvars(TTTemp, Date, Hours,'Before','Temperature'); % Add 'Hours' & 'Date' Variables
TTTemp.Properties.VariableNames(1:2) = {'Date','Hours'}; % Name 'Hours' & 'Date' Variables
TTTempT = timetable2table(TTTemp); % Convert To 'table'
VarPerHourMonthT{k,:} = unstack(TTTempT(:,2:end),'Temperature','Hours', 'VariableNamingRule','preserve'); % Unstack & Write To Cell Array
end
VarPerHourMonthT % Display Results
VarPerHourMonthT{1}(1:5,:)
VarPerHourMonthT{12}(1:5,:)
The new ‘VarPerHourMonthT’ cell array is of table arrays, not timetable arrays because I doubt that timetable arrays would support this format. The essential function here is the unstack function, and most of the code in the loop iterations is devoted to creating table arrays that are compatible with it. Using unstack on the timetable arrays themselves is possible (I did that first), however they do not produce the result you want, so I adopted this approach instead. (The problem with using unstack on the timetable arrays themselves is that it produces a diagonal matrix of ‘Temperature’ sums in a matrix of NaN values for each hour, rather than a single row. To get the result you want required creating both the ‘Date’ and ‘Hour’ variables, converting the timetable arrays to table arrays, eliminating the ‘date_time’ variable, and then using unstack on the table arrays.)
I did my best to comment-document my code here, along with describing it.
.
Ancalagon8
on 16 Jan 2023
Star Strider
on 16 Jan 2023
My pleasure!
My code works in R2022b. Since you apparently have a different (earlier) version/release (that I may not have access to even if I knew what it was), you may have to experiment with my code to get the same result. For the time being, eliminate that part of my unstack call and hope for the best! (It may be necessary to use the ‘VarPerHourMonthT{k,:}.Properties.VariableNames’ or ‘VarPerHourMonthT.Properties.VariableNames’ option to set their names. I am hesitant to experiment with that with my code since it works as written.)
Upgrading to R2022b would be the easiest option if that is available to you.
Star Strider
on 16 Jan 2023
As always, my pleasure!
Yes.
See the documentation section on Spreadsheet Files Only . It will probably be necesary to use a for loop to write to each sheet. (I have never had to do this, so I have no experience with it.)
Testing it here —
LD = load(websave('dataset','https://www.mathworks.com/matlabcentral/answers/uploaded_files/1255052/dataset.mat'));
T = LD.TT1;
VarPerHour = retime(T, 'hourly', 'sum')
for k = 1:12
MMidx = month(VarPerHour.date_time) == k;
VarPerHourMonth{k,:} = VarPerHour(MMidx,:);
end
for k = 1:12
TTTemp = VarPerHourMonth{k}; % Create Temporary 'timetable'
Hours = hour(TTTemp.date_time); % Create 'Hours' Variable
[y,m,d] = ymd(TTTemp.date_time); % Begin To Create 'Date' Variable
Date = datetime(y,m,d); % Finish Creating 'Date' Variable
TTTemp = addvars(TTTemp, Date, Hours,'Before','Temperature'); % Add 'Hours' & 'Date' Variables
TTTemp.Properties.VariableNames(1:2) = {'Date','Hours'}; % Name 'Hours' & 'Date' Variables
TTTempT = timetable2table(TTTemp); % Convert To 'table'
VarPerHourMonthT{k,:} = unstack(TTTempT(:,2:end),'Temperature','Hours', 'VariableNamingRule','preserve'); % Unstack & Write To Cell Array
end
Filename = 'RainPerHourMonth.xlsx';
for k = 1:12
writetable(VarPerHourMonthT{k}, Filename, 'Sheet',k)
end
T6 = readtable(Filename, 'Sheet',6, 'VariableNamingRule','preserve') % Check 6
I do not understand the writetable warnings. It seems to work.
.
Star Strider
on 17 Jan 2023
In your edit to your previous Comment, did anything change that requires a specific reply? (To the best of my knowledge, I do not have the table you are referring to, so I am using the one I have, and the code I created previously to work with it.)
Ancalagon8
on 17 Jan 2023
Star Strider
on 17 Jan 2023
Thank you!
Let’s do that experiment —
LD = load(websave('dataset','https://www.mathworks.com/matlabcentral/answers/uploaded_files/1255052/dataset.mat'));
T = LD.TT1;
VarPerHour = retime(T, 'hourly', 'sum')
for k = 1:12
MMidx = month(VarPerHour.date_time) == k;
VarPerHourMonth{k,:} = VarPerHour(MMidx,:);
end
for k = 1:12
TTTemp = VarPerHourMonth{k}; % Create Temporary 'timetable'
Hours = hour(TTTemp.date_time); % Create 'Hours' Variable
[y,m,d] = ymd(TTTemp.date_time); % Begin To Create 'Date' Variable
Date = datetime(y,m,d); % Finish Creating 'Date' Variable
TTTemp = addvars(TTTemp, Date, Hours,'Before','Temperature'); % Add 'Hours' & 'Date' Variables
TTTemp.Properties.VariableNames(1:2) = {'Date','Hours'}; % Name 'Hours' & 'Date' Variables
TTTempT = timetable2table(TTTemp); % Convert To 'table'
VarPerHourMonthT{k,:} = unstack(TTTempT(:,2:end),'Temperature','Hours', 'VariableNamingRule','preserve'); % Unstack & Write To Cell Array
MMM{k,:} = month(TTTemp.date_time(1,:),'shortname');
end
Filename = 'RainPerHourMonth.xlsx';
for k = 1:12
writetable(VarPerHourMonthT{k}, Filename, 'Sheet',string(MMM{k}))
end
T6 = readtable(Filename, 'Sheet','Jun', 'VariableNamingRule','preserve') % Check 6
For some reason, writetable doesn’t like the cell array (even though indexing into it should produce a character array), however it accepts the string argument. Referring to it by name in the readtable test works. (The ‘MMM’ cell array didn’t initially appear in this version of my code, so I added it in the loop.)
So, an emphatic ‘Yes!’
.
Star Strider
on 24 Jan 2023
As always, my pleasure!
More Answers (1)
Christopher McCausland
on 15 Jan 2023
Hi Ancalogon,
Time Step
'yearly'
'quarterly'
'monthly'
'weekly'
'daily'
'hourly'
'minutely'
'secondly'
Have you tried;
VarPermonth = retime(T, 'monthly', 'sum');
Let me know if this is what you are looking for, if not please provide a snippit of the data and the expected output.
Christopher
1 Comment
Christopher McCausland
on 15 Jan 2023
Hi Ancalagon,
I get what you want now.
What you really need to do is filter ValPerHour by months, heres an example of how to do so:
And also an ealier suggestion from Walter;
I hope this helps!
Christopher
Categories
Find more on Timetables in Help Center and File Exchange
Community Treasure Hunt
Find the treasures in MATLAB Central and discover how the community can help you!
Start Hunting!
