How to make sparsity pattern graph with grouping

Hi. I would like to create a graph looking like the following:

The rows correspond to periods (months), and the columns to a set of features/predictors/X's. The features have been assigned to a category, which corresponds to the different colors.

The data that I would like to plot something similar for, are in the following form (-assuming we have 30 periods, and 15 perdictors):

%Create a Sparse Matrix M:
T=30; N=15;
M=zeros(T,N);
for i=1:T
    M(i,randsample(1:N, 6, false))=1;
end
TT = array2timetable(M,'RowTimes',dateshift(datetime('today'),'end','month',-T+1:0),'VariableNames',"x"+string(1:N))

TT = 30x15 timetable
       Time        x1    x2    x3    x4    x5    x6    x7    x8    x9    x10    x11    x12    x13    x14    x15
    ___________    __    __    __    __    __    __    __    __    __    ___    ___    ___    ___    ___    ___

    30-Nov-2021    1     0     0     1     1     1     0     0     0      1      0      0      0      0      1 
    31-Dec-2021    0     0     1     1     0     0     0     1     1      0      0      1      1      0      0 
    31-Jan-2022    1     1     0     0     1     0     0     0     0      0      0      0      1      1      1 
    28-Feb-2022    0     0     0     1     0     0     1     0     0      0      1      1      1      1      0 
    31-Mar-2022    0     0     0     1     0     0     0     0     1      0      0      1      1      1      1 
    30-Apr-2022    1     0     1     0     1     1     0     0     1      1      0      0      0      0      0 
    31-May-2022    1     0     0     1     0     1     1     1     0      0      0      0      0      0      1 
    30-Jun-2022    1     0     0     0     1     0     1     0     1      0      0      0      1      0      1 
    31-Jul-2022    0     0     0     0     1     0     0     0     0      1      1      1      1      1      0 
    31-Aug-2022    1     0     0     1     1     0     0     0     0      1      1      0      0      0      1 
    30-Sep-2022    1     0     0     0     1     0     0     1     0      0      0      1      0      1      1 
    31-Oct-2022    0     1     1     0     0     0     0     1     0      1      0      0      1      1      0 
    30-Nov-2022    0     1     1     0     1     0     1     1     0      0      1      0      0      0      0 
    31-Dec-2022    0     0     1     1     0     0     1     1     1      0      1      0      0      0      0 
    31-Jan-2023    0     1     0     0     1     0     1     1     0      1      0      0      0      0      1 
    28-Feb-2023    0     0     1     1     1     1     0     0     1      1      0      0      0      0      0 

c = [repmat("Financial",[1,5]) repmat("Econ",[1,8]) repmat("Survey",[1,2])]';
c = c(randsample(1:N,N,false));
categoriez = array2table(c,'RowNames',"x"+string(1:N),"VariableNames","categ")

categoriez = 15x1 table
              categ   
           ___________

    x1     "Econ"     
    x2     "Survey"   
    x3     "Econ"     
    x4     "Financial"
    x5     "Econ"     
    x6     "Survey"   
    x7     "Financial"
    x8     "Econ"     
    x9     "Econ"     
    x10    "Econ"     
    x11    "Financial"
    x12    "Financial"
    x13    "Financial"
    x14    "Econ"     
    x15    "Econ"     

I would like to create a graph like the one on the top, using the data in timetable TT and table categoriez. Any help would be appreciated!

Answer 1

[Adam Danz]

Using pcolor

This solution uses pcolor and converts the table of 0s and 1s into grouping values that are used to assign color to each cell.

OP's demo code

T=30; N=15;
M=zeros(T,N);
for i=1:T
    M(i,randsample(1:N, 6, false))=1;
end
TT = array2timetable(M,'RowTimes',dateshift(datetime('today'),'end','month',-T+1:0),'VariableNames',"x"+string(1:N));
c = [repmat("Financial",[1,5]) repmat("Econ",[1,8]) repmat("Survey",[1,2])]';
c = c(randsample(1:N,N,false));
categoriez = array2table(c,'RowNames',"x"+string(1:N),"VariableNames","categ");

% Convert strings to categorical
categoriez.categ = categorical(categoriez.categ);

% Convert the categoricals to integers for grouping
catVals = double(categoriez.categ);

% Replace the 0s in TT with NaNs and the 1s with the grouping values.
m = TT{:,:}; 
m(m==0) = nan;
m = m.* catVals.';
m(:,end+1) = nan; % pad 1 column so last column of data is shown
m(end+1,:) = nan; % pad 1 row so last row of data is shown

% Plot the results.  
h = pcolor(1:width(m),[TT.Properties.RowTimes;NaT],m);

% Set color
ncats = numel(unique(catVals));
colormap(jet(ncats))  % nx3 matrix where n = number of categories

% Show colorbar with centered ticks
cb = colorbar();
cb.Ticks = linspace(1,ncats,ncats+1) + (ncats-1)/(ncats*2);  
cb.TickLabels = categories(categoriez.categ);

Using scatter

Another approach is to use scatter which, unlike line objects, lets you control the color of each marking by setting the cdata.

% Find the row and column coordinates of 1 values.
[row,col] = find(TT{:,:}); 

% Compute the color values (m)
% Convert strings to categorical
categoriez.categ = categorical(categoriez.categ);

% Convert the categoricals to integers for grouping
catVals = double(categoriez.categ);

% Replace the 0s in TT with NaNs and the 1s with the grouping values.
m = TT{:,:}; 
m(m==0) = nan;
m = m.* catVals.';

% plot the scatter with filled markers
scatter(col,TT.Properties.RowTimes(row),60,m(~isnan(m)),'filled')
% Set color
ncats = numel(unique(catVals));
colormap(jet(ncats))  % nx3 matrix where n = number of categories

% Show colorbar with centered ticks
cb = colorbar();
cb.Ticks = linspace(1,ncats,ncats+1) + (ncats-1)/(ncats*2);  
cb.TickLabels = categories(categoriez.categ);

[Haris K.]

That's very good indeed. Thank you Adam!

In the graph I provided, they sort the X's wrt the categories which results in colors being grouped together. I can adjust the solution you gave by sorting the timetable and the table:

[~,idx] = sort(categoriez.categ)
categoriez = categoriez(idx,:)
TT = TT(:,idx)

However, it seems that the graph in the picture I provided is using a line-plot with markers, and then probably they set the line color to white, leaving only the markers. I'll wait a bit to see if someone can came up with that solution, which seems a bit more condensed. In reality, I have 280 predictors, 220 periods, and 12 catergories.

[Adam Danz]

I recommend using scatter instead of line. With scatter, you can control the marker color for each marker. I updated my answer to show how that would look.

[Haris K.]

It now looks even better. Is there any alternative to the colorbar for showing the categories?

[Adam Danz]

You could use a legend instead but you'll need to use a hack to create the legend objects.

Add this to your code following the scatter example. I did not test this, check that the order of colors and labels are correct.

colors = jet(ncats); 
names = categories(categoriez.categ); 
hold on
h = gobjects(1,ncats); 
for i = 1:height(colors)
    h(i) = plot(nan,nan,'o',...
        'MarkerEdgeColor','none',...
        'MarkerFaceColor',colors(i,:),...
        'MarkerSize',8, ...
        'DisplayName', names{i});
end
legend(h)