Adding vertical trendline for mutiple y variables in one scatterplot

Hi,

I have a scatterplot with mutiple y variables from different trials. I want to make a trendline for all the trials together, not separate trendlines for each trial. I was reccomended to "draw" a vertical line on the x axis, then all of the trial points that touch that line should be averaged and a single point drawn on that x plane for the accumulative trendline. Then, to make the trendline, I would have to connect all these new dots together from each point of the x plane. Does this make sense? I'm not sure how to do it and would appreciate help!

Thanks!

Answer 1

[Star Strider]

One option is to simply reshape all of the different ‘x’ and ‘y’ data vectors (combined into matrices here for convenience) into a single column using either reshape or the ‘(:)’ operator, and then perform the regression —

x = sort(randn(20, 5));
y = 5*(1:size(x,2)) + randn(size(x)) + 0.5*(1:size(x,1)).';

Line = [x(:) ones(size(x(:)))] * ([x(:) ones(size(x(:)))] \ y(:));              % Calculate Single Linear Regression Line For All Data

figure
hs = scatter(x, y, 'filled', 'DisplayName','Data');
hold on
hp = plot(x(:), Line, 'DisplayName','Linear Regression');
hold off
grid
xlabel('X')
ylabel('Y')
legend([hs hp], 'Location','best')

That is how I would approach it, anyway.

.

[Sia Sharma]

I'm getting an error on this section:

Line = [x(:) ones(size(x(:)))] * ([x(:) ones(size(x(:)))] \ y(:)); % Calculate Single Linear Regression Line For All Data

• it says "Subscripting into a table using one subscript (as in t(i)) is not supported. Specify a row subscript and a variable subscript, as in t(rows,vars). To select variables, use t(:,i) or for one variable t.(i). To select rows, use t(i,:)"

This is what i set my x and y as:

x = data(:,1)

y = data(:,2:9)

[Star Strider]

That is certainly true for table arrays.

There are two options, the first is to use the table2array function to get ‘data’ as an array rather than a table. You can then use the addressing in the code you quoted.

The second is to change the addressing to use curly brackets {} instead, for example:

x = data{:,1}

y = data{:,2:9}

since that will also extract the data from the table, creating arrays.

It would likely be easier to use the second (curly brackets) option.

NOTE — In my original code, both ‘x’ and ‘y’ are matrices, so reshaping them into vectors produces vectors of equal lengths. Your ‘x’ is a vector (with I assume the same number of rows as ‘y’), so you will need to do:

x = repmat(x, size(y,2), 1)

This will make their row lengths equal, and the rest of my code should run without error.

.

[Sia Sharma]

Hi,

This almost worked but I think the statement "x = repmat(x, size(y,2), 1)" made the x rows increase much more than my y rows.

x in workspace is now 2464x1 double

y in workspace is 308x8 double

Both should be 308 rows long. Do you know how to fix this? I think the statement made my x rows increase, not my x columns.

[Star Strider]

That should work, since my ‘Line’ variable reshapes ‘y’ to create it as a column vector that will be the same size as ‘x’. It will work with the ‘x’ I calculated earlier using repmat.

Try it!

[Sia Sharma]

Hi,

It gives me an error that says :

Error using scatter

X and Y must be vectors of the same length, matrices of the same size, or a combination of a vector and a matrix

where the length of the vector matches either the number of rows or columns of the matrix.

Please help!!

[Star Strider]

I do not understand the problem. My code should work if your data matches my assumptions of it.

I need your data and your code to see what the problem is.

[Sia Sharma]

My data is a excel file with x variable seconds, y variables mmHg values for different trials. All trials go up to 301 seconds, and I have 8 trials.

My code is below. Thanks for your help!

x = data{:,1}

y = data{:,2:9}

x = repmat(x, size(y,2), 1)

Line = [x(:) ones(size(x(:)))] * ([x(:) ones(size(x(:)))] \ y(:)); % Calculate Single Linear Regression Line For All Data

figure

hs = scatter(x, y, 'filled', 'DisplayName','Data');

hold on

hp = plot(x(:), Line, 'DisplayName','Linear Regression');

hold off

grid

xlabel('X')

ylabel('Y')

legend([hs hp], 'Location','best')

[Star Strider]

My pleasure!

I now see the problem in adapting my code to your data.

Try this —

data = array2table(randn(308,9).*(1:9)+(0:8)*10);                   % Create 'data'

xv = data{:,1};                                                     % Change This To 'xv'
y = data{:,2:9};
x = repmat(xv, size(y,2), 1);                                       % Change Reference To 'xv' Here
Line = [x(:) ones(size(x(:)))] * ([x(:) ones(size(x(:)))] \ y(:));  % Calculate Single Linear Regression Line For All Data

figure
hs = scatter(xv, y, 'filled', 'DisplayName','Data');
hold on
hp = plot(x(:), Line, '-k', 'LineWidth',2, 'DisplayName','Linear Regression');
hold off
grid
xlabel('X')
ylabel('Y')
legend([hs hp], 'Location','best')

.

[Sia Sharma]

It worked! Thank you so much 😊

[Star Strider]

As always, my pleasure!