MimModels                package:mimR                R Documentation

_C_r_e_a_t_e _u_n_d_i_r_e_c_t_e_d _a_n_d _b_l_o_c_k _r_e_c_u_r_s_i_v_e _M_I_M _m_o_d_e_l_s

_D_e_s_c_r_i_p_t_i_o_n:

     ...........

_U_s_a_g_e:

     mim(mimFormula, data, letter=FALSE, marginal=data$name)

_A_r_g_u_m_e_n_t_s:

mimFormula: A model formula following the MIM syntax. Long variable
          names are allowed however. See 'details'. The formula can be
          given either with a tilde or as a string

    data: A gmData object

  letter: If TRUE, the variables used in mim.formula are single
          letters.

marginal: Can be used for specifying only a subset of the variables in
          connection with a main effects, a saturated and a homogeneous
          saturated model

_D_e_t_a_i_l_s:

     A mim.formula can be
     "Sex+Drug/Sex:W1+Drug:W1+Sex:W2+Drug:W2/Sex:W1:W2+Drug:W1:W2" or
     (if letter is TRUE) the shorter form  "ab/abx,aby/abxy"  or
     "ab/abx+aby/abxy". A mimFormula can also be "." (the main effects
     (the independence) model), ".." (the saturated model) or "..h"
     (the homogeneous saturated model). See 'examples'.

_V_a_l_u_e:

     A mimModel or mimBRModel object

_N_o_t_e:

     Before using mimR, make sure that the MIM program is runnning.

_A_u_t_h_o_r(_s):

     Sren Hjsgaard, sorenh@agrsci.dk

_R_e_f_e_r_e_n_c_e_s:

     David Edwards, An Introduction to Graphical Modelling, Springer
     Verlag, 2002

_S_e_e _A_l_s_o:

     'as.gmData'

_E_x_a_m_p_l_e_s:

     # Create som models (no data needed!)
     gmd.rats.nodata  <-  gmData(c("Sex","Drug","W1","W2"),
         factor=c(2,3,FALSE,FALSE),
         vallabels=list("Sex"=c("M","F"), "Drug"=c("D1","D2","D3")))

     m12   <- mim("Sex:Drug/Sex:Drug:W1+Sex:Drug:W2/W1:W2", data=gmd.rats.nodata)
     m22   <- mim("ab/abc+abd/cd", data=gmd.rats.nodata, letter=TRUE)

     summary(m12)
     summary(m22)

     m.main <- mim(".",  data=gmd.rats.nodata)
     m.sat  <- mim("..",  data=gmd.rats.nodata)
     m.hsat <- mim("..h", data=gmd.rats.nodata)

     summary(m.main); 
     summary(m.sat); 
     summary(m.hsat)

     # Next we need some data to work with
     data(rats)
     gmd.rats <- as.gmData(rats)
     vallabels(gmd.rats)
     observations(gmd.rats)

     m1   <- mim("Sex:Drug/Sex:Drug:W1+Sex:Drug:W2/W1:W2", data=gmd.rats)
     m2   <- mim("ab/abc+abd/cd", data=gmd.rats, letter=TRUE)

     m.main <- mim(".",   data=gmd.rats, marginal=c("Sex", "Drug", "W1"))
     m.sat  <- mim("..",  data=gmd.rats, marginal=c("Sex", "Drug", "W1"))
     m.hsat <- mim("..h", data=gmd.rats, marginal=c("Sex", "Drug", "W1"))

     m1f  <- fit(m1)
     m2f  <- fit(m2)

     summary(m1f)
     summary(m2f)

     m.main <- fit(mim(".",  data=gmd.rats))
     m.sat  <- fit(mim("..",  data=gmd.rats))
     m.hsat <- fit(mim("..h", data=gmd.rats))

     summary(m.main); 
     summary(m.sat); 
     summary(m.hsat)

     # To generate an nth order hierarchical log-linear model for discrete
     # data you can do

     data(HairEyeColor)
     mim(nthOrderModel(names(dimnames(HairEyeColor)),order=2),data=as.gmData(HairEyeColor))

