Prev: Reading large file and sorting
Next: please can any one post the matlab script for these s•generate 100 random number or use the test data provided by the standard;• write Convolution Encoder using MATLAB script; Then you need to verifythe Convolution Encoder by • call
From: Artur Stoppa on 19 Jan 2010 14:33
I have a code :

% testowanie bazy glass.data
% glass.data - http://archive.ics.uci.edu/ml/
Data = load('glass.data', '-ascii');
Data(:,1) = [];
[NumOfRows, NumOfCols] = size (Data);
Classes = Data(:,NumOfCols);
Data(:,NumOfCols) = [];
MainSet = dataset(Data,Classes);
clear Data;


MAXT = 95; % maksymalna liczba klasyfikatorów
TSTART = 5;
TSTEP = 10;
ITERATIONS = 10;


Result = [];
VoteResult = [];
AvgVoteResult = zeros((MAXT + TSTART) / TSTEP, 2);

% podział na zbiór uczący i testowy

[LearnSet, TestSet, LI, TI] = gendat(MainSet,0.8);
[NumOfRows, NumOfCols] = size (LearnSet);
[tsNumOfClasses, tmp] = size(getlablist(TestSet));

Learning = LearnSet * treec;
Testing = TestSet * Learning;
Error1 = Testing * testc;
Error1



for iter = 1:ITERATIONS

VoteResult = [];
for j = TSTART:TSTEP:MAXT

LearnedClassifiers = cell (1,1);
k = 0;
for i = 1:j
% utwórz zbiór uczący (losujac ze zwracaniem z LearnSet)
NewSetIndices = betarnd(2,2,NumOfRows,1);
NewSetIndices = round(NumOfRows.*NewSetIndices);
NewSetIndices(find (NewSetIndices == 0)) = 1;
NewSetIndices(find (NewSetIndices > NumOfRows)) = NumOfRows;
NewSetLabels = getlab(LearnSet);
NewLearnSet = dataset(LearnSet.data(NewSetIndices,:),NewSetLabels(NewSetIndices));
% w całym procesie biorą udział zbiory o jednakowej liczbie klas
[NumOfClasses, tmp] = size(getlablist(NewLearnSet));
if (NumOfClasses == tsNumOfClasses)
k = k + 1;
% LearnSets{1,i} = NewLearnSet; %zagregowanie zbiorów uczących w cell array (opcjonalne)

% Uczymy klasyfikator i dodajemy do zbioru
LearningTmp = NewLearnSet * treec;
LearnedClassifiers{1,k} = LearningTmp;%NewLearnSet * treec;

% badanie jakości klasyfikatora wylosowanego (stopa błędu)
%TestingTmp = TestSet * LearningTmp;
%ErrorT = TestingTmp * testc;
%ResultVector = [j, i, ErrorT];
%Result = [Result; [j, i, ErrorT]];
end;
% ewentualne uwzględnianie jakości klasyfikatora przy wyborze do
% komitetu
% if 0.7*ErrorT > Error1
%sprintf('Błąd klasyfikatora %d większy od błędu klasyfikatora pierwotnego\nNie bierze udziału w głosowaniu', i)
% else
% VotingLearnSets {1,i} = NewLearnSet;
%sprintf('AAAAAAAAAAAAAAAAAAA')
% end;
end;

% testowanie decyzji grupowej komitetu (klasyfikatory kombinowane
% votec)
VALL = [LearnedClassifiers{:}];
VC = VALL * votec;
Testing = TestSet * VC;
ErVote = Testing * testc;
% zebranie wartości błędów w macierz
VoteResult = [VoteResult; [j, ErVote]];
end;

AvgVoteResult(:,2) = AvgVoteResult(:,2) + VoteResult(:,2);
%AvgVoteResult = [AvgVoteResult VoteResult(:,2)];
end;
AvgVoteResult(:,1) = VoteResult(:,1);
AvgVoteResult(:,2) = AvgVoteResult(:,2) / ITERATIONS;

when i run that , I only see :
"
PR_Warning: getprior: testc: No priors found in dataset, class frequencies are used instead
PR_Warning: getprior: testc: No priors found in dataset, class frequencies are used instead
PR_Warning: getprior: testc: No priors found in dataset, class frequencies are used instead
PR_Warning: getprior: testc: No priors found in dataset, class frequencies are used instead
"
I have this program from my friend - he said to me that he worked.
I have to do : "The quality of classifiers, after the passage through bugging, set against with the quality of primitive classifiers"
 | 
Pages: 1
Prev: Reading large file and sorting
Next: please can any one post the matlab script for these s•generate 100 random number or use the test data provided by the standard;• write Convolution Encoder using MATLAB script; Then you need to verifythe Convolution Encoder by • call