Scatterplot arrays with different number of elements and tracing the trendline

So... I have 3 arrays, one is called "M_IRI" (1 row and 36 columns) and the other is called "EVAN_DSUV_FSDN_100" (3 rows and 36 columns) and "RESTO_100" (3 rows and 36 columns) respectively . My problem is happening when I try to make a scatterplot... It just doesn't work.

I need to make this scatterplot to get the trend line between these values, one of my colleagues said to do it separately and then just plot the average of these 2 3x36 matrices, but statistically, this is wrong, so I need to somehow get these 2 trend lines, one between M_IRI and EVAN_DSUV_FSDN_100 and between M_IRI and RESTO_100.

Is there any way to do this?

[Jon]

It isn't at all clear to me what you are trying to do from your description. Exactly which variable do you want plotted in the y-axis (vertical) of your scatter plot(s) and which variables do you want as the x-axis (horizontal).

For example if you wanted to plot the second row of EVAN_DSUV_FSDN_100 (y-axis) vs M_IRI (x-axis), you could use

plot(M_IRI,EVAN_DSUV_FSDN_100(2,:),'o')
xlabel('M_IRI')
ylabel('EVAN_DSUV_FSDN_100 ')

You could also use

scatter(M_IRI,EVAN_DSUV_FSDN_100(2,:))
xlabel('M_IRI')
ylabel('EVAN_DSUV_FSDN_100 ')

[Jon]

Reading your question more carefully, and in particular your last sentence I think I understand what you are trying to do. I would recommend using @Star Strider's answer

Answer 1

[Star Strider]

One option is to reshape the (3x36) arrays into (i08x1) (column) vectors and duplicate ‘M_IRI’ to match by just triplicating it as a column vector. (I chose column vectors because this simple linear regression requires them.) Then do the regression.

Try this —

M_IRI = rand(1,36);
EVAN_DSUV_FSDN_100 = randn(3, 36)+[0;1;2];
RESTO_100 = randn(3, 36)+[5;6;7];

B1 = [repmat(M_IRI(:), 3, 1) ones(36*3,1)] \ reshape(EVAN_DSUV_FSDN_100.', [], 1)

B1 = 2×1

0.1971 0.8247

B2 = [repmat(M_IRI(:), 3, 1) ones(36*3,1)] \ reshape(RESTO_100.', [], 1)

B2 = 2×1

-0.2154 6.1815

figure
scatter(M_IRI, EVAN_DSUV_FSDN_100)
hold on
plot([min(M_IRI) max(M_IRI)], [min(M_IRI) 1; max(M_IRI) 1]*B1, '-r')
hold off

figure
scatter(M_IRI, RESTO_100)
hold on
plot([min(M_IRI) max(M_IRI)], [min(M_IRI) 1; max(M_IRI) 1]*B2, '-r')
hold off

These are just random data so the trends are not impressive, however this should work with your data. You can also use polyfit and polyval to do the regression.

EDIT — Corrected typographical errors.

.