Increasing the number of iterations in Generalize​dLinearMod​el.fit

Dear Experts,

I am fitting a matrix of predictors (desmat) to a timeseries (ts) and to do it I use a GeneralizedLinearModel object as follows:

m = GeneralizedLinearModel.fit(desmat, ts);

However, I often get the following warning:

Warning: Iteration limit reached. 
> In glmfit (line 332)
  In GeneralizedLinearModel/postFit (line 605)
  In classreg.regr.FitObject/doFit (line 95)
  In GeneralizedLinearModel.fit (line 887)

I have to use GeneralizedLinearModel.fit instead of glmfit because I am also running some contrasts between the model parameters down the line.

My question is this: how can I increase the number of maximum iterations in this fit method, so I can try to make my model converge? Alternatively, is there a way to do a contrast of coefficients that returns p-values using the glmfit function?

Thank you very much,

Leonardo Tozzi

Answer 1

[Aditya Patil]

As per my understanding, you want to get the p values from the fitted model. You can use fitglm for this purpose. You can increase the iterations using the MaxIter option.

load hospital
dsa = hospital;
modelspec = 'Smoker ~ Age*Weight*Sex - Age:Weight:Sex';
mdl = fitglm(dsa,modelspec,'Distribution','binomial','Options',statset('MaxIter',1000))

mdl = 
Generalized linear regression model:
    logit(Smoker) ~ 1 + Sex*Age + Sex*Weight + Age*Weight
    Distribution = Binomial

Estimated Coefficients:
                        Estimate         SE         tStat      pValue 
                       ___________    _________    ________    _______

    (Intercept)            -6.0492       19.749     -0.3063    0.75938
    Sex_Male               -2.2859       12.424    -0.18399    0.85402
    Age                    0.11691      0.50977     0.22934    0.81861
    Weight                0.031109      0.15208     0.20455    0.83792
    Sex_Male:Age          0.020734      0.20681     0.10025    0.92014
    Sex_Male:Weight        0.01216     0.053168     0.22871     0.8191
    Age:Weight         -0.00071959    0.0038964    -0.18468    0.85348


100 observations, 93 error degrees of freedom
Dispersion: 1
Chi^2-statistic vs. constant model: 5.07, p-value = 0.535

[Leonardo Tozzi]

Unfortunately, the p-values I would like to get are for the difference of two effects estimates, not of the effects themselves. Sorry if this was not clear. My code to achieve this at the moment is as follows:

m = GeneralizedLinearModel.fit(desmat, ts);
con=zeros(1, height(m.Coefficients));
con(2)=1;
con(3)=-1;
[p_glm(v),f_glm(v)]=coefTest(m,con);