3D surface plot in linear regression model

Hi, I used fitlm to obtain linear regression of an experiment in which response Mass is related to H, T and V. I obtained the model with reasonable R-squares and p-value. So I assume the regression model is fine. Now I would like to plot a 3D surface, visualise the response surface of Mass(z-axis) related to H(x-axis) and V(y-axis). All I can find in Matlab tutorial is plotsice function that seems relevant but what I want is a 3D surface. Anyone knows how to do this? Thank you.

Answer 1

[John D'Errico]

No. You have a FOUR dimensional thing. This is not a surface in 3 dimensions. It is a surface in FOUR dimensions, thus Mass as a function of three variables, (H,T,V).

If you fix the 4th variable that you seem to be ignoring (T) at some value, then use meshgrid to generate points in a grid of H and V. Then evaluate the model at all those points, using surf to plot the result. The result will be valid for the given value of T as chosen in advance.

[Pingfei Jiang]

I see, so I will need to fix one of the predictors, e.g. T = 1, then use the model to calculate the response of in a grid of H and V (24x24), then use surf to plot, correct?

[John D'Errico]

Exactly.

[Pingfei Jiang]

Great, thank you very much, i will accept your answer