run regression, change the value of x and calculate how y changes, and then plot both

(Revised) Suppose I have a table T

y x1 x2
10 3  4
7  2  3
5  1  2

run a regression

mdl =fitlm([T.X1, T.X2],T.y)

Then I now want to calculate T.y1 for T.x3= T.x1-1 using the above fitlm formula

y x1 x2 x3
10 3  4 2
7  2  3 1 
5  1  2 0 

Answer 1

[Image Analyst]

Use a for loop to alter T.X1

for numRuns = 1 : 10 % However many you want
    x = [T.X1, T.X2];
    mdl =fitlm(x, T.y)
    % Plot stuff...
    plot(x, T.y, '-');
    grid on;
    hold on;
    % Now do something with mdl, like get T.y1.
    % Now decrement T.X1 by 1 and do the next run.
    T.X1 = T.X1 - 1;
end

[alpedhuez]

Please see the revision.

[Image Analyst]

You're still making us work too much to help you. Give code to create T please.

fitlm() takes a table directly but it expects y to be the last column, not the first.

mdl = fitlm(tbl) returns a linear regression model fit to variables in the table or dataset array tbl. By default, fitlm takes the last variable as the response variable.

So you'd need

T2 = [T(:, 2:3), T.y]; % Rearrange columns so x1 and x2 are first, then y is last.
mdl = fitlm(T2)

[alpedhuez]

T is just some data.

The idea is "I estimate the model of Y given X1 and X2. Now suppose we reduce X1 by 1. Then what new Y will look like?"

[Image Analyst]

Wow, that's helpful.

Maybe I'll get to it someday then.

So, did you try to adapt the for loop where I reduced X1 by 1 on each iteration? If not, why not?

[alpedhuez]

https://www.mathworks.com/help/stats/linearmodel.predict.html