9804bad1775b284d22e7423375df4a63364548f4,mlxtend/classifier/adaline.py,Adaline,fit,#Adaline#,62

Before Change

        // initialize weights
        if init_weights:
            self._init_weights(shape=1 + X.shape[1])

        self.cost_ = []

        if self.minibatches is None:
            self.w_ = self._normal_equation(X, y)

        // Gradient descent or stochastic gradient descent learning
        else:
            n_idx = list(range(y.shape[0]))
            self.init_time_ = time()
            for i in range(self.epochs):
                if self.minibatches > 1:
                    X, y = self._shuffle(X, y)

                minis = np.array_split(n_idx, self.minibatches)
                for idx in minis:
                    y_val = self.activation(X[idx])
                    errors = (y[idx] - y_val)
                    self.w_[1:] += self.eta * X[idx].T.dot(errors)
                    self.w_[0] += self.eta * errors.sum()

                cost = self._sum_squared_error_cost(y, self.activation(X))
                self.cost_.append(cost)

After Change

            self.thres_ = 0.5

        if init_weights:
            self.w_ = self._init_weights(shape=1 + X.shape[1],
                                         zero_init_weight=self.zero_init_weight,
                                         seed=self.random_seed)

        self.cost_ = []

        if self.minibatches is None:
            self.w_ = self._normal_equation(X, y)

        // Gradient descent or stochastic gradient descent learning
        else:
            n_idx = list(range(y.shape[0]))
            self.init_time_ = time()
            for i in range(self.epochs):
                if self.minibatches > 1:
                    X, y = self._shuffle(X, y)

                minis = np.array_split(n_idx, self.minibatches)
                for idx in minis:
                    y_val = self.activation(X[idx])
                    errors = (y[idx] - y_val)
                    self.w_[1:] += self.eta * X[idx].T.dot(errors)
                    self.w_[0] += self.eta * errors.sum()

                cost = self._sum_squared_error_cost(y, self.activation(X))
                self.cost_.append(cost)

Instances