Timetable Variable grouped by year and stackedplot

I have the following unstacked timetable that I would like to plot with the stackedplot function.

The stackedplot should have the Day/Month as the X-Axis and it should be "stacked" or grouped by the year.

Should the grouping be done by some timetable function or should I just loop through the timetable with the required conditions? Thanks!

    Datum       DailyAvgTemp
    __________    ____________

    01.01.1956        -6.8    
    02.01.1956        -3.7    
    03.01.1956        -4.5    
    04.01.1956        -5.2    
    05.01.1956           0    
    06.01.1956        -3.8 
    ...
    05.11.2018          -2    
    06.11.2018        -2.8 

Answer 1

[Adam Danz]

Steps taken to break apart a timetable by years and plot in stackedplot

1. The timetable is broken apart by year using arrayfun and each subtable is stored in a cell array TTbyYear.
2. The dates of each subtable are shifted to 1952, preserving the month and day. Note that 1952 is a leapyear. It's important that all dates share the same year and since some years in your data may be leapyears, the shared year should be a leapyear.
3. The subtables are joined back together using outerjoin where each year has its own column and the variable names in the new table TTy are the original years.
4. The new table TTy is sent to stackedplot().

Demo

% Create demo timetable
TT = timetable(round(rand(3652,1)*100)/10-5,...
     'StartTime',datetime(1956,1,1,'Format','dd.MM.yyyy'),...
     'TimeStep',days(1),...
     'VariableNames',{'DailyAvgTemp'});
TT.Properties.DimensionNames{1} = 'Datum';

head(TT) % show sample

ans = 8x1 timetable
      Datum       DailyAvgTemp
    __________    ____________

    01.01.1956        -1.1    
    02.01.1956          -3    
    03.01.1956         2.4    
    04.01.1956         2.4    
    05.01.1956        -4.2    
    06.01.1956        -1.8    
    07.01.1956          -4    
    08.01.1956        -4.4    

% Break apart timetable by year
TTyears = year(TT.Datum);
unqYears = unique(TTyears);
TTbyYear = arrayfun(@(y){TT(TTyears==y,:)},unqYears);

% Join the tables back into a timetable with 1 column for each year
baseYear = 1956; % should be a leap year
TTy = timetable();
for i = 1:numel(TTbyYear)
    % Change year to base year
    TTbyYear{i}.Datum = datetime(baseYear,month(TTbyYear{i}.Datum),day(TTbyYear{i}.Datum));
    % join tables
    TTy = outerjoin(TTy, TTbyYear{i}); 
end
% Change variable names to years
TTy.Properties.VariableNames = compose('%d',unqYears);
% Change format of datetime values to hide the year
TTy.Datum.Format = 'dd.MM';

head(TTy) % Show sample

ans = 8x10 timetable
    Datum    1956    1957    1958    1959    1960    1961    1962    1963    1964    1965
    _____    ____    ____    ____    ____    ____    ____    ____    ____    ____    ____

    01.01    -1.1    -0.6     2.4     2.3    -2.7    -1.7      -3     4.8     2.2      -2
    02.01      -3    -3.1    -0.1     4.4    -0.4     0.1    -3.3    -2.3     2.2     3.9
    03.01     2.4     4.6     0.3    -2.1     0.8    -2.3     2.9     1.8    -2.5    -3.5
    04.01     2.4    -4.6    -0.1       1    -3.4     1.1    -2.8     1.9       3       4
    05.01    -4.2    -2.6    -2.5    -1.4     4.4     2.1     3.6    -2.2    -2.1     4.5
    06.01    -1.8    -4.4    -3.7     4.9       0    -2.9     4.5    -0.9    -0.6     4.4
    07.01      -4     3.8    -3.4    -4.6       1    -0.2     2.8     2.6     2.5      -1
    08.01    -4.4    -0.6     2.1    -0.4     1.8     0.6    -0.4     1.3    -3.1    -3.9

% Plot
stackedplot(TTy)

●

[John Barron]

Thank you Sir!

This answered all my questions!

[Adam Danz]

Glad I could help.

I just made a small change to the demo. The variable baseYear was not used and I hard-coded the year within the loop instead. That was fixed so that baseYear is used within the loop.

Answer 2

[Cris LaPierre]

For stackedplot to work as you intend, you would have to make each year's data its own variable (column). There is no timetable function for that. Also note that stackedplot has a maximum of 25 variables (1 variable = 1 plot). It looks like you might have 63.

I do think turning your table into a timetable, or at the least making your first column datetimes will make it much easier to work with the time data.

[John Barron]

Thank you for your reply!

The dataset is already a timetable.

Would you suggest using some other plotting method or would you really create new variable as "years" ?

[Cris LaPierre]

You don't need it, and year(TT.date) can extract it easily enough.

While not ideal if there are 63 years of data, one way to show this in a single plot would be this.

% create a sample dataset
datum = (datetime(1956,1,1):calmonths(1):datetime(1960,12,1))';
avgTemp = randi(100,[length(datum),1]);
TT = timetable(datum,avgTemp)

TT = 60x1 timetable
       datum       avgTemp
    ___________    _______

    01-Jan-1956      28   
    01-Feb-1956      49   
    01-Mar-1956      27   
    01-Apr-1956      64   
    01-May-1956      12   
    01-Jun-1956      55   
    01-Jul-1956      34   
    01-Aug-1956      51   
    01-Sep-1956      12   
    01-Oct-1956      20   
    01-Nov-1956      89   
    01-Dec-1956      69   
    01-Jan-1957       9   
    01-Feb-1957      14   
    01-Mar-1957       7   
    01-Apr-1957      63   

% Plot grouping by year
h=gscatter(month(TT.datum),TT.avgTemp,year(TT.datum));
set(h,'LineStyle','-');
xticks(1:12);
xticklabels(["Jan" "Feb" "Mar" "Apr" "May" "Jun" "Jul" "Aug" "Sep" "Oct" "Nov" "Dec"]);
legend('Location','eastoutside')

[Adam Danz]

If you were set on using a stacked plot, you could break apart your data into 3-5 year segments and show multiple years in each axes.

[Cris LaPierre]

Looks like I incorrectly assumed dates were mm-dd-yyyy, which is simpler to solve. If it's dd-mm-yyyy, it gets a little more challenging.