naivebayes.m

%% NAIVE BAYES (TEST included)

%% LOAD DATA
setup()
[content, cat, ~] = loadData('data/data.csv',1);

%%
% Gerar matrix de treino e de teste
ptreino = 0.7;
[TRAIN, TEST] = trainAndTestData(content, cat, ptreino);

%%
% Calcular as prob de ser de cada categoria
[PCAT, UNIQCAT] = probCat(cat);

%%
% Calcular as prob de ser de catx sabendo que tem as caracteristicas y
[PWORD, relevantWords] = probWordGivenCat(TRAIN,UNIQCAT);

%% PREDICTIONS

nTestEntries = size(TEST,1);

prediction = cell(1,nTestEntries);
for n = 1:nTestEntries
    elem = TEST{n,1};
    res = classifyNB(elem, PWORD, PCAT, relevantWords);
    [~, argmax] = max(res);
    prediction{n} = UNIQCAT{argmax};
end

real = TEST(:,2);

%% CONFUSION MATRIX
cm = confusionmat(real,prediction);
ch = confusionchart(cm, UNIQCAT,...
    "OffDiagonalColor",[0.3010 0.7450 0.9330]);

tp = cm(1,1);
tn = cm(2,2);
fp = cm(1,2);
fn = cm(2,1);

a = (tp + tn) / (tp + tn + fp + fn);
r = (tp) / (tp + fn);
p = (tp) / (tp + fp);
f1 = (2 * p * r) / (p + r);
fprintf("Accuracy: %.3f\n", a);
fprintf("Recall: %.3f\n", r);
fprintf("Precision: %.3f\n", p);
fprintf("F1 measure: %.3f\n", f1);

%% SAVING DATA

saveNB(PWORD, relevantWords, PCAT, UNIQCAT);
clear