DataMethods              package:rmutil              R Documentation

_M_e_t_h_o_d_s _f_o_r _r_e_s_p_o_n_s_e, _t_c_c_o_v, _t_v_c_o_v, _a_n_d _r_e_p_e_a_t_e_d _D_a_t_a _O_b_j_e_c_t_s

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

     Objects of class, 'response', contain response values, and
     possibly the corresponding times, binomial totals, nesting
     categories, censor indicators, and/or units of precision/Jacobian.
     Objects of class, 'tccov', contain time-constant or
     inter-individual, baseline covariates. Objects of class, 'tvcov',
     contain time-varying or intra-individual covariates. Objects of
     class, 'repeated', contain a 'response' object and possibly
     'tccov' and 'tvcov' objects.

     In formula and functions, the key words, 'times' can be used to
     refer to the response times from the data object as a covariate,
     'individuals' to the index for individuals as a factor covariate,
     and 'nesting' the index for nesting as a factor covariate. The
     latter two only work for W&R notation.

     The following methods are available for accessing the contents of
     such data objects.

     'as.data.frame': places all of the variables in the data object in
     one dataframe, extending time-constant covariates to the length of
     the others unless the object has class, 'tccov'. Binomial and
     censored response variables have two columns, respectively `yes'
     and `no' and response and censoring indicator, with the name given
     to the response.

     'as.matrix': places all of the variables in the data object in one
     matrix, extending time-constant covariates to the length of the
     others unless the object has class, 'tccov'. If any covariates are
     factor variables (instead of the corresponding sets of indicator
     variables), the matrix will be character instead of numeric.

     'covariates': extracts covariate matrices from a data object (for
     formulae and functions, possibly for selected individuals. See
     'covariates.formulafn').

     'covind': gives the indexing of the response by individual (that
     is, the nesting indicator for observations within individuals). It
     can be used to expand time-constant covariates to the size of the
     repeated measurements response.

     'delta': extracts the units of measurement vector and Jacobian of
     any transformation of the response, possibly for selected
     individuals. Note that, if the unit of measurement/Jacobian is
     available in the 'response' object, this is automatically included
     in the calculation of the likelihood function in all library model
     functions.

     'units': prints the variable names and their description and
     returns the latter.

     'formula': gives the formula used to create the time-constant
     covariate matrix of a data object (for formulae and functions, see
     'formula.formulafn').

     'names': extracts the names of the response and/or covariates.

     'nesting': gives the coding variable(s) for individuals (same as
     'covind') and also for nesting within individuals if available,
     possibly for selected individuals.

     'nobs': gives the number of observations per individual.

     'plot': plots the variables in the data object in various ways.
     For 'repeated' objects, 'name' can be a response or a time-varying
     covariate.

     'print': prints summary information about the variables in a data
     object.

     'response': extracts the response vector, possibly for selected
     individuals. If there are censored observations, this is a
     two-column matrix, with the censor indicator in the second column.
     For binomial data, it is a two-column matrix with "positive" (y)
     and "negative" (totals-y) frequencies.

     'resptype': extracts the type of each response.

     'times': extracts the times vector, possibly for selected
     individuals.

     'transform': transforms variables. For example, 'transform(z,
     y=fcn1(y), times=fcn2(times))' where 'fcn1' and 'fcn2' are
     transformation functions. When the response is transformed, the
     Jacobian is automatically calculated. New response variables and
     covariates can be created in this way, if the left hand side is a
     new name ('ynew=fcn3(y)'), as well as replacing an old variable
     with the transformed one. If the transformation reverses the order
     of the responses, use its negative to keep the ordering and have a
     positive Jacobian; for example, 'ry=-1/y'. For 'repeated' objects,
     only the response and the times can be transformed.

     'units': prints the variable names and their units of measurement
     and returns the latter.

     'weights': extracts the weight vector, possibly for selected
     individuals.

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

     as.data.frame(z)
     as.matrix(z)
     covariates(z, nind=NULL, names=NULL, expand=F)
     covind(z)
     delta(z, nind=NULL, names=NULL)
     formula(z)
     names(z)
     nesting(z, nind=NULL)
     nobs(z)
     plot(z, name=NULL, nind=NULL, nest=1, ccov=NULL, add=F, lty=NULL, pch=NULL,
             main=NULL, ylim=NULL, xlim=NULL, xlab=NULL, ylab=NULL, ...)
     print(z, nindmax=50)
     response(z, nind=NULL, names=NULL)
     resptype(z)
     times(z, nind=NULL)
     transform(z, times=NULL, ...)
     units(z)
     weights(z, nind=NULL)

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

       z: A 'response', 'tccov', 'tvcov', or 'repeated' data object.
          For 'covind' and 'nobs', this may also be a model.

       y: The function, when the response is to be transformed.

   times: The function, when the times are to be transformed.

   names: The names of the response variable(s) or covariate(s).

    nind: The numbers of individuals to be used. (For plotting, cannot
          be used simultaneously with 'ccov'.)

  expand: For intra-class (time-constant) covariates, if TRUE, expand
          them to give one value per observation rather than one per
          individual. Only works with 'repeated' objects when all
          individuals are requested ('nind=NULL').

    ccov: For plotting: If a vector of values for the time-constant
          covariates is supplied, only individuals having that set of
          values will have profiles plotted. These values must be in
          the order in which the covariates appear when the data object
          is printed. For factor variables, the codes must be given. If
          the name of a covariate is supplied, a set of graphs is
          plotted, one for each covariate value, showing profiles of
          all individuals having that value. (The covariate can have a
          maximum of six values.) Cannot be used simultaneously with
          'nind'.

    nest: For plotting: nesting category to plot.

     add: For plotting: add to previous plot.

 nindmax: For printing a 'response', 'tvcov', or 'repeated' object, if
          the number of individuals is greater than 'nindmax', the
          range of numbers of observations per individual is printed
          instead of the vector of numbers.

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

     These methods extract information stored in 'response', 'tccov',
     'tvcov', and 'repeated' data objects created respectively by
     'restovec', 'tcctomat', 'tvctomat', and 'rmna'.

     Note that if a vector of binomial totals or a censoring indicator
     is present, this is extract as the second column of the matrix
     produced by the 'response' method.

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

     J.K. Lindsey

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

     'restovec', 'rmna', 'tcctomat', 'tvctomat'.

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

     # set up some data and create the objects
     #
     y <- matrix(rnorm(20),ncol=5)
     tt <- c(1,3,6,10,15)
     print(resp <- restovec(y, times=tt, units="m", type="duration"))
     x <- c(0,0,1,1)
     x2 <- as.factor(c("a","b","a","b"))
     tcc <- tcctomat(data.frame(x=x,x2=x2))
     z <- matrix(rpois(20,5),ncol=5)
     tvc <- tvctomat(z)
     print(reps <- rmna(resp, tvcov=tvc, ccov=tcc))
     #
     plot(resp)
     plot(reps)
     plot(reps, nind=1:2)
     plot(reps, ccov=c(0,1))
     plot(reps, ccov="x2")
     plot(reps, name="z", nind=3:4, pch=1:2)
     plot(reps, name="z", ccov="x2")
     #
     response(resp)
     response(transform(resp, y=1/y))
     response(reps)
     response(reps, nind=2:3)
     response(transform(reps,y=1/y))
     #
     times(resp)
     times(transform(resp,times=times-6))
     times(reps)
     #
     delta(resp)
     delta(reps)
     delta(transform(reps,y=1/y))
     delta(transform(reps,y=1/y), nind=3)
     #
     nobs(resp)
     nobs(tvc)
     nobs(reps)
     #
     units(resp)
     units(reps)
     #
     resptype(resp)
     resptype(reps)
     #
     weights(resp)
     weights(reps)
     #
     covariates(tcc)
     covariates(tcc, nind=2:3)
     covariates(tvc)
     covariates(tvc, nind=3)
     covariates(reps)
     covariates(reps, nind=3)
     covariates(reps,names="x")
     covariates(reps,names="z")
     #
     names(tcc)
     names(tvc)
     names(reps)
     #
     nesting(resp)
     nesting(reps)
     #
     # because individuals are the only nesting, this is the same as
     covind(resp)
     covind(reps)
     #
     as.data.frame(resp)
     as.data.frame(tcc)
     as.data.frame(tvc)
     as.data.frame(reps)
     #
     # use in glm
     rm(y,x,z)
     glm(y~x+z, data=reps)

