5 examples of 'roc curve sklearn' in Python

499
def test_roc(self):
500
    self.lr_roc_curve(self.max_lr_f1(True))

109
def plot_roc(y_test, y_pred, label=''):
110
    """Compute ROC curve and ROC area"""
111
112
    fpr, tpr, _ = roc_curve(y_test, y_pred)
113
    roc_auc = auc(fpr, tpr)
114
115
    # Plot of a ROC curve for a specific class
116
    plt.figure()
117
    plt.plot(fpr, tpr, label='ROC curve (area = %0.2f)' % roc_auc)
118
    plt.plot([0, 1], [0, 1], 'k--')
119
    plt.xlim([0.0, 1.0])
120
    plt.ylim([0.0, 1.05])
121
    plt.xlabel('False Positive Rate')
122
    plt.ylabel('True Positive Rate')
123
    plt.title('Receiver operating characteristic' + label)
124
    plt.legend(loc="lower right")
125
    plt.show()

302
def get_roc_curve(model, data, thread_count=-1, plot=False):
303
    """
304
    Build points of ROC curve.
305
306
    Parameters
307
    ----------
308
    model : catboost.CatBoost
309
        The trained model.
310
311
    data : catboost.Pool or list of catboost.Pool
312
        A set of samples to build ROC curve with.
313
314
    thread_count : int (default=-1)
315
        Number of threads to work with.
316
        If -1, then the number of threads is set to the number of CPU cores.
317
318
    plot : bool, optional (default=False)
319
        If True, draw curve.
320
321
    Returns
322
    -------
323
    curve points : tuple of three arrays (fpr, tpr, thresholds)
324
    """
325
    if type(data) == Pool:
326
        data = [data]
327
    if not isinstance(data, list):
328
        raise CatBoostError('data must be a catboost.Pool or list of pools.')
329
    for pool in data:
330
        if not isinstance(pool, Pool):
331
            raise CatBoostError('one of data pools is not catboost.Pool')
332
333
    roc_curve = _get_roc_curve(model._object, data, thread_count)
334
335
    if plot:
336
        with _import_matplotlib() as plt:
337
            _draw(plt, roc_curve[0], roc_curve[1], 'False Positive Rate', 'True Positive Rate', 'ROC Curve')
338
339
    return roc_curve

21
def ROC_plot(features,X_,y_, pred_,title):
22
    fpr_, tpr_, thresholds = roc_curve(y_, pred_)
23
    optimal_idx = np.argmax(tpr_ - fpr_)
24
#https://stackoverflow.com/questions/28719067/roc-curve-and-cut-off-point-python
25
    optimal_threshold = thresholds[optimal_idx]
26
    auc_ = auc(fpr_, tpr_)
27
    title = "{} auc=".format(title)
28
    print("{} auc={} OT={:.4g}".format(title, auc_,optimal_threshold))
29
    plt.plot(fpr_, tpr_, label="{}:{:.4g}".format(title, auc_))
30
    plt.xlabel('False positive rate')
31
    plt.ylabel('True positive rate')
32
    plt.title('SMPLEs={} Features={} OT={:.4g}'.format(X_.shape[0],len(features),optimal_threshold))
33
    plt.legend(loc='best')
34
    plt.savefig("./_auc_[{}].jpg".format(features))
35
    plt.show()
36
    return auc_,optimal_threshold

220
def precision_recall_curve(clf, x_test, y_test):
221
    from sklearn.metrics import precision_recall_curve
222
    
223
    for i in range(2):
224
        y_probabilities = [x[i] for x in clf.predict_proba(x_test)]
225
        precision, recall, thresholds = precision_recall_curve(y_test, y_probabilities)
226
227
        plt.title('Precision Recall Curve')
228
        plt.plot(recall, precision, 'b')
229
230
    plt.show()