Method _test

pattern/vector/__init__.py:1799–1822

Returns an (accuracy, precision, recall, F1-score)-tuple for the given documents, with values between 0.0 and 1.0 (0-100%). Accuracy is the percentage of correct predictions for the given test set, but this metric can be misleading (e.g., classifier *always* pred

(self, documents=[], target=None, **kwargs)

        return cls(train=d[:i]).test(d[i:])
    
    def _test(self, documents=[], target=None, **kwargs):
        """ Returns an (accuracy, precision, recall, F1-score)-tuple for the given documents,
            with values between 0.0 and 1.0 (0-100%).
            Accuracy is the percentage of correct predictions for the given test set,
            but this metric can be misleading (e.g., classifier *always* predicts True).
            Precision is the percentage of predictions that were correct.
            Recall is the percentage of documents that were correctly labeled.
            F1-score is the harmonic mean of precision and recall.
        """
        A = [] # Accuracy.
        P = [] # Precision.
        R = [] # Recall.
        for type, TP, TN, FP, FN in self.confusion_matrix(documents).split():
            if type == target or target is None:
                # Calculate precision & recall per class.
                A.append(float(TP + TN) / ((TP + TN + FP + FN)))
                P.append(float(TP) / ((TP + FP) or 1))
                R.append(float(TP) / ((TP + FN) or 1))
        # Macro-averaged:
        A = sum(A) / (len(A) or 1)
        P = sum(P) / (len(P) or 1)
        R = sum(R) / (len(R) or 1)
        F = 2.0 * P * R / ((P + R) or 1)
        return A, P, R, F

    def confusion_matrix(self, documents=[]):
        """ Returns the confusion matrix for the given test data,