85d02c39457ee57f5545fbea6079c518a5fec3b8,src/skmultiflow/meta/adaptive_random_forest_regressor.py,ARFRegBaseLearner,_normalize_drift_input,#ARFRegBaseLearner#Any#,476

Before Change

        drift_input = drift_input[0]
        if drift_input < self._min_drift_data:
            self._min_drift_data = drift_input
        if drift_input > self._max_drift_data:
            self._max_drift_data = drift_input

        if self._min_drift_data != self._max_drift_data:
            return (drift_input - self._min_drift_data) / \
                (self._max_drift_data - self._min_drift_data)
        else:

After Change

        self._k += 1
        // Welford"s algorithm update step
        if self._k == 1:
            self._pM = self._M = drift_input
            self._pS = 0

            return 0.0
        else:
            self._M = self._pM + (drift_input - self._pM) / self._k
            self._S = self._pS + (drift_input - self._pM) * (drift_input - self._M)

            // Save previously calculated values for the next iteration
            self._pM = self._M
            self._pS = self._S

            sd = np.sqrt(self._S / (self._k - 1))

            // Apply z-score normalization to drift input
            norm_input = (drift_input - self._M) / sd

            // Data with zero mean and unit variance -> (empirical rule) 99.73% of the values lie
            // between [mean - 3*sd, mean + 3*sd] (in a normal distribution): we assume this range
            // for the norm variable.
            // Hence, the values are assumed to be between [-3, 3] and we can apply the min-max norm
            // to cope with ADWIN"s requeriments
            return (norm_input + 3) / 6

    def reset(self, instances_seen):
        if self._use_background_learner and self.background_learner is not None:
            self.estimator = self.background_learner.estimator
            self.evaluator = self.background_learner.evaluator

Instances