function [a_db, varargout] = mean(a_db, dim)
% mean - Returns the mean of the data matrix of a_db. Ignores NaN values.
%
% Usage:
% [a_db, n] = mean(a_db, dim)
%
% Description:
% Does a recursive operation over dimensions in order to remove NaN values.
% This takes more time than a straightforward mean operation.
%
% Parameters:
% a_db: A tests_db object.
% dim: Work down dimension.
%
% Returns:
% a_db: The DB with one row of mean values.
% n: (Optional) Numbers of non-NaN rows included in calculating each column.
%
% See also: mean, tests_db
%
% $Id$
%
% Author: Cengiz Gunay <cgunay@emory.edu>, 2004/10/06
% Copyright (c) 2007 Cengiz Gunay <cengique@users.sf.net>.
% This work is licensed under the Academic Free License ("AFL")
% v. 3.0. To view a copy of this license, please look at the COPYING
% file distributed with this software or visit
% http://opensource.org/licenses/afl-3.0.php.
if ~ exist('dim', 'var')
dim = 1; % Go down rows by default
end
% Always do row-wise
order = 1:length(dbsize(a_db));
if dim ~= 1
order(dim) = 1;
order(1) = dim;
data = permute(a_db.data, order);
else
data = a_db.data;
end
% Allocate results array
db_size = size(data);
s = repmat(NaN, [1 db_size(2:end)]);
% Do a loop over EACH other dimension (!)
[s, n] = recmean(data, length(db_size));
if dim ~= 1
s = ipermute(s, order);
end
a_db = set(a_db, 'id', [ 'Mean of ' get(a_db, 'id') ]);
a_db = set(a_db, 'data', s);
nout = max(nargout,1) - 1;
if nout > 0
varargout{1} = n;
end
% Recursive std needed for stripping NaNs in each dimension
function [s, n] = recmean(data, dim)
if dim == 1
sdata = data(~isnan(data(:)) & ~isinf(data(:)));
n = size(sdata, 1);
if n == 0
% If a divide by zero error occured,
% give it NaN value instead of an empty matrix.
s = NaN;
else
s = mean(sdata, 1);
end
else
for num=1:size(data, dim)
% Otherwise recurse
[dims{1:(dim-1)}] = deal(':');
dims{dim} = num;
[s(dims{:}) n(dims{:})] = recmean(data(dims{:}), dim - 1);
end
end