source: prdatasets/prmissingvalues.m @ 86

function [x,msg] = prmissingvalues(x,val)
%PRMISSINGVALUES Fix the missing values in a dataset
%
%          [X,MSG] = PRMISSINGVALUES(X,VAL)
%
% Fix the missing values (represented by NaN's) in dataset X. String MSG
% gives a text message of what has been done.
%
% The following values VAL are possible:
%          'remove'     remove entries that contain missing values
%          'mean'       fill the entries with the mean of their 
%                       respective column
%          <value>      fill the entries with a fixed constant
%
% See also datasets

% Copyright: D.M.J. Tax, D.M.J.Tax@prtools.org
% Faculty EWI, Delft University of Technology
% P.O. Box 5031, 2600 GA Delft, The Netherlands

% Where are the offenders?
I = isnan(x);

% If there are missing values, go:
if any(I(:))
        switch val
        case {'remove' 'delete'}
                I = find(sum(I,2)==0);
                x = x(I,:);
                msg = 'Entries with missing values have been removed.';
        case 'mean'
                k = size(x,2);
                for i=1:k
                        J = ~I(:,i);
                        if any(I(:,i)) %is there a missing value in this feature?
                                if ~any(J)
                                        error('Missing value cannot be filled: all values are NaN.');
                                end
                                mn = mean(x(J,i));
                                x(find(I(:,i)),i) = mn;
                        end
                end
                msg = 'Missing values have been replaced by the mean of the feature.';
        otherwise
                if ~isa(val,'double')
                        error('Missing values can only be filled by scalars.');
                end
                x(I) = val;
                msg = sprintf('Missing values have been replaced by %f.',val);
        end
end
return