determinewinner.m

有监督自组织映射-偏最小二乘算法(A supervised self-organising map–partial least squares algorithm),可以用语多变量数据的回归分析

function [Value, Pos] = DetermineWinner(X,Map,DistType,MLKP);

DistType=upper(DistType);
if DistType == 'EUC'
% calculates minimum of Euclidean distances
    Distance = dist(X,Map);
    [Value, Pos] = min(Distance);
elseif DistType == 'ABS'
% calculates minimum of sum of absolute values
    Distance=sum(abs(Map-(ones(MLKP.Nunits,1)*X)'));
    [Value, Pos] = min(Distance);
elseif DistType == 'TAN'
% calculates maximum of Tanimoto distances
% tricky: usually, many Tanimoto distances are at minimum
    Distance = TanimotoDistance(X,Map,MLKP);
    [Value, Pos] = min(Distance);
elseif DistType == 'WCC'
    CorrValues=WccValues(X,Map,MLKP);
    [Value, Pos] = max(CorrValues);
    Distance = ones(1,MLKP.Nunits) - CorrValues;
elseif DistType == 'WDD'
    Distance=WddValues(X,Map,MLKP);
    [Value, Pos] = min(Distance);
else
% calculates maximum of correlations
    CorrValues = CalcCorrValues(X,Map);
    [Value, Pos] = max(CorrValues);
    Distance = ones(1,MLKP.Nunits) - CorrValues;
end
Distance = Distance/max(Distance);
%
% if necessary, calculate normalised inner-product for weighting
if (upper(MLKP.AdaptLearn) == 'Y' | upper(MLKP.BoltzmannMode) ~= 'NOT')
    Value = 2 - min(Distance);
else
    Value = 1;
end