How to write an automatic function based on the median of data to form groups of data.

How to write an automatic function based on the median of data to form groups of data.

as the boxplots are here in this figure, i need to make groups or clusters of data based on median. like once i have median of data, then i need to identify the beginning of a new cluster/group when the median delay increase instead of decresing. and when the median again increases then from that point a new group or cluster will start.

i manually grouped the data based on median, i need to make function like this.

Thanks in advance.

Answer 1

[Adam Danz]

% Create demo data : nxm matrix with 1 box per column
rng('default')
data = rand(100,15) + [1.0 1.3 0.8 0.95 1.02 1.05 1.08 1.2 1.0 1.03 1.06 0.90 0.95 0.80 1.1];

% Compute median of each column
dataMedians = median(data);

% Create boxplot
h = boxplot(data);

% Add group separation lines
groupStart = diff([inf,dataMedians])<0;
groupIdx = find(groupStart);
for i = 1:numel(groupIdx)
    xline(groupIdx(i)-.5, 'k--', 'LineWidth',1)
end

% Color the groups 
colors = lines(sum(groupStart));
groupID = cumsum(groupStart);
for i = 1:sum(groupStart)
    set(h(:,groupID==i), 'Color', colors(i,:))
end

Using boxchart with group separation lines (for color control see demo1, demo2, documentation demo.

% Create demo data : nxm matrix with 1 box per column
rng('default')
data = rand(100,15) + [1.0 1.3 0.8 0.95 1.02 1.05 1.08 1.2 1.0 1.03 1.06 0.90 0.95 0.80 1.1];

% Compute median of each column
dataMedians = median(data);

% Create boxplot
figure()
h = boxchart(data);
% Add group separation lines
groupStart = diff([inf,dataMedians])<0;
groupIdx = find(groupStart);
for i = 1:numel(groupIdx)
    xline(groupIdx(i)-.5, 'k--', 'LineWidth',1)
end

[zhoug zho]

Thank you so much Adam Danz .