SOM Toolbox Online documentation


csS = som_stats(D,varargin)

SOM_STATS Calculate descriptive statistics for the data.
 csS = som_stats(D,[sort]); 
  csS = som_stats(D); 
  csS = som_stats(D,'nosort'); 

  Input and output arguments ([]'s are optional): 
   D           (matrix) a matrix, size dlen x dim
               (struct) data or map struct
   [sort]      (string) 'sort' (default) or 'nosort'
                        If 'nosort' is specified, the data is not 
                        sorted, and therefore the values of
                        nunique, uvalues, ucount, fvalues, fcount, and tiles fields 
                        are not calculated. This may be useful if
                        there is a very large amount of data, and
                        one wants to reduce calculation time.

   csS         (cell array) size dim x 1, of statistics structs with 
                        the following fields
      .type             (string) 'som_stat'
      .name             (string) name of the variable
      .normalization    (struct array) variable normalization (see SOM_NORMALIZE)
      .ntotal           (scalar) total number of values
      .nvalid           (scalar) number of valid values (not Inf or NaN)
      .min              (scalar) minimum value 
      .max              (scalar) maximum value 
      .mean             (scalar) mean value (not Inf or NaN)
      .std              (scalar) standard deviation (not Inf or NaN)
      .nunique          (scalar) number of unique values
      .mfvalue          (vector) most frequent value
      .mfcount          (vector) number of occurances of most frequent value
      .values           (vector) at most MAXDISCRETE (see below) sample values 
      .counts           (vector) number of occurances for each sampled value
      .tiles            (vector) NT-tile values, for example
                                    NT=4   for quartiles: 25%, 50% and 75%
                                    NT=100 for percentiles: 1%, 2%, ... and 99%
      .hist             (struct) histogram struct with the following fields
           .type        (string) 'som_hist'
           .bins        (vector) histogram bin centers 
           .counts      (vector) count of values in each bin
           .binlabels   (cellstr) labels for the bins (denormalized bin
                                  center values)
           .binlabels2  (cellstr) labels for the bins (denormalized bin
                                  edge values, e.g. '[1.4,2.5['

      MAXDISCRETE = 10
      NT          = 10


[ SOM Toolbox online doc ]