glm - generalized linear model function with Scilab
glm function is the main function of the glmbox Scilab package produced to analyse some of most popular statistical models belonging to wide class of the generalized linear models (GLM) such as Binomial models, Poisson models, Normal models, Gamma models and Inverse Gaussian models.
[Result] = glm(Y,X,distribution,link_function,dispersion_parameter,constant,residual_type,maxiter,accuracy) [Result] = glm(Y,X,distribution)
The first version of glm function provides more control for the input parameters; the second version of glm function provides a set of default parameters which are usually used for the distribution chosen
for Gamma distribution the default parameters
link_function=reciprocal_link   dispersion_parameter='mean_deviance' constant=1 residual_type='pearson' maxiter=1000 accuracy=0.001
for Inverse Gaussian distribution the default parameters
link_function=power_link(-2)   dispersion_parameter='mean_deviance' constant=1 residual_type='pearson' maxiter=1000 accuracy=0.001
for Normal distribution the default parameters
link_function=identity_link   dispersion_parameter='mean_deviance' constant=1 residual_type='pearson' maxiter=1000 accuracy=0.001
for Poisson distribution the default parameters
link_function=logarithmic_link   dispersion_parameter=1 constant=1 residual_type='pearson' maxiter=1000 accuracy=0.001
for Binomial distribution the default parameters
link_function=logit_link   dispersion_parameter=1 constant=1 residual_type='pearson' maxiter=1000 accuracy=0.001
References:
[1] Dobson, A.J. (1990), "An Introduction to Generalized Linear Models", CRC Press.
[2] Gill, J. (2001), "Generalized Linear Models: a unified approach", (quantitative applications in the social sciences, 134). Sage Pubblications.
[3] McCullagh, P., and J.A. Nelder (1990), "Generalized Linear Models", CRC Press.
[4] Myers, R.M., Montgomery, D.C. and Vining, G.G. (2001), "Generalized Linear Models with applications in engineering and the science", John Wiley and Sons.