perClass Documentation
version 5.1 (31-May-2017)
SDMISSING  Show, remove or replace missing values

 Show info about missing data in data set
    SDMISSING(DATA)

 Remove samples with missing values
    CLEAN=SDMISSING(DATA)
    [CLEAN,MIS,INDCL,INDMIS]=SDMISSING(DATA)

 Replace missing values (Imputation)
    NEW=SDMISSING(DATA,options)

 INPUT
    DATA     SDDATA set

 OUTPUT
    CLEAN    SDDATA set with samples without missing values
    MIS      SDDATA ser with samples containint missing values
    INDCL    indices of CLEAN samples in DATA
    INDMIS   indices of MIS samples in DATA
    NEW      new SDDATA object of same size as DATA with filled in missing values

 OPTIONS
   'mean','median' replace each missing value by a mean/median of all
          existing data for the same feature
   'class mean','class median' replace missing values by a mean/median of
          existing data in the same class (this is only useful in training)
   'from',ORIG  replace missing values in DATA based on existing values in ORIG

 EXAMPLES
 Replace NaNs in A by median of non-NaN values in A
   B=sdmissing(A,'median')
 Replace NaNs by median of a respective class, take values from the data set C:
   B=sdmissing(A,'class median','from',C)