rangenorm               package:dprep               R Documentation

_r_a_n_g_e _n_o_r_m_a_l_i_z_a_t_i_o_n

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

     Performs several methods of range normalization

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

     rangenorm(data, method = c("znorm", "mmnorm", "decscale", 
     "signorm", "softmaxnorm"),superv=TRUE)

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

    data: the name of the dataset to be normalized

  method: the discretization method to be used:"znorm", "mmnrom",
          "decscale", "signorm", "softmaxnorm"

  superv: 

     {superv=T for supervised data, that data including the class
     labels in the last column.  if superv=F means that the data to be
     used is unsupervised.}

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

     In the znorm  normalization, the mean of each attribute of the
     transformed set of data points is reduced to zero by subtracting
     the mean of each attribute from the values of the attributes and
     dividing the result by the standard deviation of the attribute.
     Uses the function scale found in the base library.

     Min-max normalization (mmnorm) subtracts the minimum value of an
     attribute from each value of the attribute and then divides the
     difference by the range of the attribute. These new values are
     multiplied by the new range of the attribute and finally added to
     the new minimum value of the attribute. These operations transform
     the data into a new range, generally [0,1].

     The decscale normalization applies decimal scaling to a matrix or
     dataframe. Decimal scaling transforms the data into [-1,1] by
     finding k such that  the absolute value of the maximum value of
     each attribute divided by 10\^k   is less than or equal to 1.

     In the sigmoidal normalization (signorm) the input data is
     nonlinearly  transformed into [-1,1] using a sigmoid function. The
     original data is first  centered about the mean, and then mapped
     to the almost linear region of the sigmoid. Is especially
     appropriate when outlying values are present.

     The softmax normalization is so called because it reaches "softly"
     towards maximum and minimum value, never quite getting there. The
     transformation is more or less linear in the middle range, and has
     a nonlinearity at both ends. The output range covered is [0,1]. 
     The algorithm removes the classes of the dataset before
     normalization and replaces them at the end to form the matrix
     again.

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

     A matriz containing the discretized data.

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

     Caroline Rodriguez and Edgar Acuna

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

     Caroline Rodriguez (2004). An computational environmnent for data
     preprocessing  in supervised classification. Master thesis.
     UPR-Mayaguez

     Hann, J., Kamber, M. (2000). Data Mining: Concepts and Techniques.
      Morgan Kaufman Publishers.

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

     #----Several methods of range normalization ----
     data(bupa)
     bupa.znorm=rangenorm(bupa,method="znorm")
     bupa.mmnorm=rangenorm(bupa,method="mmnorm")
     bupa.decs=rangenorm(bupa,method="decscale")
     bupa.signorm=rangenorm(bupa,method="signorm")
     bupa.soft=rangenorm(bupa,method="softmaxnorm")
     #----Plotting to see the effect of the normalization----
     op=par(mfrow=c(2,3))
     plot(bupa[,1])
     plot(bupa.znorm[,1])
     plot(bupa.mmnorm[,1])
     plot(bupa.decs[,1])
     plot(bupa.signorm[,1])
     plot(bupa.soft[,1])
     par(op)

