5 examples of 'logistic regression sklearn' in Python

89
def logistic_regression(w, x):
90
    """Logistic regression classifier model.
91
92
    w: Weights w. (n_features,) NumPy array
93
    x: Data point x_i. (n_features,) NumPy array
94
    -> float in [0, 1]
95
    """
96
    return scipy.special.expit(numpy.dot(x, w.T))

94
def Logistic_Regression(X,Y,alpha,theta,num_iters):
95
	m = len(Y)
96
	for x in xrange(num_iters):
97
		new_theta = Gradient_Descent(X,Y,theta,m,alpha)
98
		theta = new_theta
99
		if x % 100 == 0:
100
			Cost_Function(X,Y,theta,m)
101
			print 'theta ', theta	
102
			print 'cost is ', Cost_Function(X,Y,theta,m)
103
	Declare_Winner(theta)

158
def logistic_regression_3(X, y, max_iter : int = 100, learning_rate : float = 0.1):
159
    W = np.zeros((np.size(X, 1), np.size(y, 1)))
160
    for _ in range(max_iter):
161
        N = len(y)
162
        index = np.random.permutation(N)
163
        X = X[index]
164
        y = y[index]
165
        W_prev = np.copy(W)
166
        y_pred = softmax(X[0:10][:] @ W)
167
        grad = X[0:10][:].T @ (y_pred - y[0:10])
168
        W -= learning_rate * grad
169
        if np.allclose(W, W_prev):
170
            break
171
    return W

1447
def predict_proba(self, X):
1448
    """
1449
    Probability estimates.
1450
1451
    The returned estimates for all classes are ordered by the
1452
    label of classes.
1453
1454
    For a multi_class problem, if multi_class is set to be "multinomial"
1455
    the softmax function is used to find the predicted probability of
1456
    each class.
1457
    Else use a one-vs-rest approach, i.e calculate the probability
1458
    of each class assuming it to be positive using the logistic function.
1459
    and normalize these values across all the classes.
1460
1461
    Parameters
1462
    ----------
1463
    X : array-like of shape (n_samples, n_features)
1464
        Vector to be scored, where `n_samples` is the number of samples and
1465
        `n_features` is the number of features.
1466
1467
    Returns
1468
    -------
1469
    T : array-like of shape (n_samples, n_classes)
1470
        Returns the probability of the sample for each class in the model,
1471
        where classes are ordered as they are in ``self.classes_``.
1472
    """
1473
    check_is_fitted(self)
1474
1475
    ovr = (self.multi_class in ["ovr", "warn"] or
1476
           (self.multi_class == 'auto' and (self.classes_.size <= 2 or
1477
                                            self.solver == 'liblinear')))
1478
    if ovr:
1479
        return super()._predict_proba_lr(X)
1480
    else:
1481
        decision = self.decision_function(X)
1482
        if decision.ndim == 1:
1483
            # Workaround for multi_class="multinomial" and binary outcomes
1484
            # which requires softmax prediction with only a 1D decision.
1485
            decision_2d = np.c_[-decision, decision]
1486
        else:
1487
            decision_2d = decision
1488
        return softmax(decision_2d, copy=False)

139
def _regress(self, x, y, alpha):
140
    kw = self.ridge_kw or {}
141
    coef = ridge_regression(x, y, alpha, **kw)
142
    self.iters += 1
143
    return coef