fdata                  package:far                  R Documentation

_F_u_n_c_t_i_o_n_a_l _D_a_t_a _c_l_a_s_s

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

     Object of class 'fdata' and its methods.

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

     as.fdata(object,...)
     as.fdata.matrix(object,..., col, p, dates, name)
     as.fdata.list(object,..., dates, name)

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

  object: A matrix or a list. 

     col: A vector giving the names of the variables to include in the
          'fdata' object. 

       p: A real value giving the number of discretization point
          chosen.

   dates: A vector of character containing the dates of the
          observations.

    name: A vector of character containing the names of the variables
          (generated if not provided). 

     ...: Additional arguments. 

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

     Fdata objects are mainly used to modelize functional data in the
     purpose of computing functional autoregressive model by the 'far'
     and 'kerfon' functions.

     An fdata is composed of one or several variables. Each ones is a
     functional time series.

     To be more precise, every variable got a functional data by
     element of the 'dates' (explicitly given or implicitly deduced).
     So the number of functional observations is a common data.

     In the contrary, each variable can be expressed in a different
     functional space. For example, if you got two variables,
     Temperature and Wind, measured during 30 days. Choosing a daily
     representation, the 'fdata' will contain a 30 elements long
     'dates' vector. Nevertheless, the variables measurement can be
     different. If Temperature is measured every hour and Wind every
     two hours, the 'fdata' object can handle such a representation.
     The only constraint is to get a regular measurement: no changes in
     the methodology.

     Basically, the 'fdata' objects are discrete measurements but the
     modelization which can be used on it will make it functional.
     Indeed, The first methods implemented as 'far' and 'kerfon' use a
     linear approximation, but more sophisticate modelization, as
     splines or wavelets approximations may come.

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

     An object of class fdata.

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

     J. Damon

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

     'far', 'multplot', 'maxfdata', 'kerfon'.

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

     # Reading of the data
     library(stats)
     data(UKDriverDeaths)

     # Making the data of class 'fdata'
     fUKDriverDeaths <- as.fdata(UKDriverDeaths,col=1,p=12,dates=1969:1984,
                                 name="UK Driver Deaths")
     summary(fUKDriverDeaths)

     # ploting of the data : whole and 1 year
     par(mfrow=c(2,1))
     plot(fUKDriverDeaths,xval=1969+(1:192)/12,whole=TRUE,
          name="Whole Evolution : ")
     plot(fUKDriverDeaths,date="1984",xval=1:12,
          name="Evolution during year 1984 : ")

     # Matrix conversion
     print(as.fdata(matrix(rnorm(50),10,5)))
     print(as.fdata(matrix(rnorm(500),100,5),col=1:2,p=5))

     # List Conversions
     print(as.fdata(list("X"=matrix(rnorm(100),10,10),
     "Z"=matrix(rnorm(50),5,10))))

