I have given some data of children: Age, weight, height.
I want to fit the following model
lm = fitlm(data,'Weight~Age+Height+Height^2+Height^3+Height^4');
By looking at the coefficients table
Estimate SE tStat pValue
__________ __________ _______ __________
(Intercept) 50.645 48.143 1.052 0.29388
Age 0.42242 0.040691 10.381 4.7207e-21
Height -6.6635 5.3439 -1.2469 0.21365
Height^2 0.37769 0.21575 1.7506 0.081312
Height^3 -0.0084367 0.0037419 -2.2547 0.025068
Height^4 6.5841e-05 2.3519e-05 2.7996 0.0055389
I see, that I might drop the variable "Height".
So far no problem, but I would like to visualize in some way, that dropping Height ist not bad or maybe is even beneficial.
Indeed matlab gives me easy acces to a visualisation, via
Since the data is 6 dimensional I wonder, how matlab does project this in 2-dim.
Especially I do not know what is meant by "Adjusted whole model".
On my search for an answer I was directed to
where it was explained, that matlab uses the "Frisch–Waugh–Lovell theorem" in some way.
But in this thread my question was not adressed, as far as I realize.
Every help would greatly be appreciated,
thank you in advance :)
