Python metrics.log_loss方法代碼示例

# 需要導入模塊: from sklearn import metrics [as 別名]
# 或者: from sklearn.metrics import log_loss [as 別名]
def multi_class_classification(data_X,data_Y):
    '''
    calculate multi-class classification and return related evaluation metrics
    '''

    svc = svm.SVC(C=1, kernel='linear')
    # X_train, X_test, y_train, y_test = train_test_split( data_X, data_Y, test_size=0.4, random_state=0) 
    clf = svc.fit(data_X, data_Y) #svm
    # array = svc.coef_
    # print array
    predicted = cross_val_predict(clf, data_X, data_Y, cv=2)
    print "accuracy",metrics.accuracy_score(data_Y, predicted)
    print "f1 score macro",metrics.f1_score(data_Y, predicted, average='macro') 
    print "f1 score micro",metrics.f1_score(data_Y, predicted, average='micro') 
    print "precision score",metrics.precision_score(data_Y, predicted, average='macro') 
    print "recall score",metrics.recall_score(data_Y, predicted, average='macro') 
    print "hamming_loss",metrics.hamming_loss(data_Y, predicted)
    print "classification_report", metrics.classification_report(data_Y, predicted)
    print "jaccard_similarity_score", metrics.jaccard_similarity_score(data_Y, predicted)
    # print "log_loss", metrics.log_loss(data_Y, predicted)
    print "zero_one_loss", metrics.zero_one_loss(data_Y, predicted)
    # print "AUC&ROC",metrics.roc_auc_score(data_Y, predicted)
    # print "matthews_corrcoef", metrics.matthews_corrcoef(data_Y, predicted) 

# 需要導入模塊: from sklearn import metrics [as 別名]
# 或者: from sklearn.metrics import log_loss [as 別名]
def evaluation_analysis(true_label,predicted): 
    '''
    return all metrics results
    '''
    print "accuracy",metrics.accuracy_score(true_label, predicted)
    print "f1 score macro",metrics.f1_score(true_label, predicted, average='macro')     
    print "f1 score micro",metrics.f1_score(true_label, predicted, average='micro') 
    print "precision score",metrics.precision_score(true_label, predicted, average='macro') 
    print "recall score",metrics.recall_score(true_label, predicted, average='macro') 
    print "hamming_loss",metrics.hamming_loss(true_label, predicted)
    print "classification_report", metrics.classification_report(true_label, predicted)
    print "jaccard_similarity_score", metrics.jaccard_similarity_score(true_label, predicted)
    print "log_loss", metrics.log_loss(true_label, predicted)
    print "zero_one_loss", metrics.zero_one_loss(true_label, predicted)
    print "AUC&ROC",metrics.roc_auc_score(true_label, predicted)
    print "matthews_corrcoef", metrics.matthews_corrcoef(true_label, predicted) 

# 需要導入模塊: from sklearn import metrics [as 別名]
# 或者: from sklearn.metrics import log_loss [as 別名]
def evaluate(val_df, clf):
    incorrect_session = {}
    val_df['scores'] = clf.predict(val_df.drop(data_drop_columns, axis=1))

    loss = log_loss(val_df.label.values, val_df.scores.values)
    grouped_val = val_df.groupby('session_id')
    rss_group = {i:[] for i in range(1,26)}
    rss = []
    for session_id, group in grouped_val:

        scores = group.scores
        sorted_arg = np.flip(np.argsort(scores))
        rss.append( group['label'].values[sorted_arg])
        rss_group[len(group)].append(group['label'].values[sorted_arg])
        if group['label'].values[sorted_arg][0] != 1:
            incorrect_session[session_id] = (sorted_arg.values, group['label'].values[sorted_arg])
    mrr = compute_mean_reciprocal_rank(rss)
    mrr_group = {i:(len(rss_group[i]), compute_mean_reciprocal_rank(rss_group[i])) for i in range(1,26)}
    print(mrr_group)
    if not configuration.debug:
        pickle.dump( incorrect_session, open(f'../output/{model_name}_val_incorrect_order.p','wb'))
    return mrr, mrr_group, loss 

# 需要導入模塊: from sklearn import metrics [as 別名]
# 或者: from sklearn.metrics import log_loss [as 別名]
def test_logreg_predict_proba_multinomial():
    X, y = make_classification(n_samples=10, n_features=20, random_state=0,
                               n_classes=3, n_informative=10)

    # Predicted probabilities using the true-entropy loss should give a
    # smaller loss than those using the ovr method.
    clf_multi = LogisticRegression(multi_class="multinomial", solver="lbfgs")
    clf_multi.fit(X, y)
    clf_multi_loss = log_loss(y, clf_multi.predict_proba(X))
    clf_ovr = LogisticRegression(multi_class="ovr", solver="lbfgs")
    clf_ovr.fit(X, y)
    clf_ovr_loss = log_loss(y, clf_ovr.predict_proba(X))
    assert_greater(clf_ovr_loss, clf_multi_loss)

    # Predicted probabilities using the soft-max function should give a
    # smaller loss than those using the logistic function.
    clf_multi_loss = log_loss(y, clf_multi.predict_proba(X))
    clf_wrong_loss = log_loss(y, clf_multi._predict_proba_lr(X))
    assert_greater(clf_wrong_loss, clf_multi_loss) 

# 需要導入模塊: from sklearn import metrics [as 別名]
# 或者: from sklearn.metrics import log_loss [as 別名]
def test_warm_start_converge_LR():
    # Test to see that the logistic regression converges on warm start,
    # with multi_class='multinomial'. Non-regressive test for #10836

    rng = np.random.RandomState(0)
    X = np.concatenate((rng.randn(100, 2) + [1, 1], rng.randn(100, 2)))
    y = np.array([1] * 100 + [-1] * 100)
    lr_no_ws = LogisticRegression(multi_class='multinomial',
                                  solver='sag', warm_start=False,
                                  random_state=0)
    lr_ws = LogisticRegression(multi_class='multinomial',
                               solver='sag', warm_start=True,
                               random_state=0)

    lr_no_ws_loss = log_loss(y, lr_no_ws.fit(X, y).predict_proba(X))
    for i in range(5):
        lr_ws.fit(X, y)
    lr_ws_loss = log_loss(y, lr_ws.predict_proba(X))
    assert_allclose(lr_no_ws_loss, lr_ws_loss, rtol=1e-5) 

# 需要導入模塊: from sklearn import metrics [as 別名]
# 或者: from sklearn.metrics import log_loss [as 別名]
def get_intercept_loss(self, model, data):
        y = np.array([x[1] for x in data.mapValues(lambda v: v.label).collect()])
        X = np.ones((len(y), 1))
        if model.model_name == 'HeteroLinearRegression' or model.model_name == 'HeteroPoissonRegression':
            intercept_model = LinearRegression(fit_intercept=False)
            trained_model = intercept_model.fit(X, y)
            pred = trained_model.predict(X)
            loss = metrics.mean_squared_error(y, pred) / 2
        elif model.model_name == 'HeteroLogisticRegression':
            intercept_model = LogisticRegression(penalty='l1', C=1e8, fit_intercept=False, solver='liblinear')
            trained_model = intercept_model.fit(X, y)
            pred = trained_model.predict(X)
            loss = metrics.log_loss(y, pred)
        else:
            raise ValueError("Unknown model received. Stepwise stopped.")
        self.intercept = intercept_model.intercept_
        return loss 

# 需要導入模塊: from sklearn import metrics [as 別名]
# 或者: from sklearn.metrics import log_loss [as 別名]
def calc_metrics(y_true, y_hat, max_steps=1000):
    y_true = np.array(y_true)
    y_hat = np.array(y_hat)
    metrics = {}
    metrics['Logloss'] = float(log_loss(y_true, y_hat))
    metrics['AUC'] = roc_auc_score(y_true, y_hat)
    metrics['F1'] = []
    metrics['Precision'] = []
    metrics['Recall'] = []
    for i in range(1, max_steps):
        threshold = float(i) / max_steps
        y_tmp = y_hat > threshold
        metrics['F1'].append(f1_score(y_true, y_tmp))
        metrics['Precision'].append(precision_score(y_true, y_tmp))
        metrics['Recall'].append(recall_score(y_true, y_tmp))
    max_idx = np.argmax(metrics['F1'])
    metrics['F1'] = metrics['F1'][max_idx]
    metrics['Precision'] = metrics['Precision'][max_idx]
    metrics['Recall'] = metrics['Recall'][max_idx]
    metrics['Threshold'] = float(max_idx + 1) / max_steps
    return metrics 

# 需要導入模塊: from sklearn import metrics [as 別名]
# 或者: from sklearn.metrics import log_loss [as 別名]
def _handle_broken_model(self, model, error):
        del model

        n = self.genome_handler.n_classes
        loss = log_loss(np.concatenate(([1], np.zeros(n - 1))), np.ones(n) / n)
        accuracy = 1 / n
        gc.collect()

        if K.backend() == 'tensorflow':
            K.clear_session()
            tf.reset_default_graph()

        print('An error occurred and the model could not train:')
        print(error)
        print(('Model assigned poor score. Please ensure that your model'
               'constraints live within your computational resources.'))
        return loss, accuracy 

# 需要導入模塊: from sklearn import metrics [as 別名]
# 或者: from sklearn.metrics import log_loss [as 別名]
def evaluate(features):
    dtrain = xgb.DMatrix(tr_x[features], label=tr_y)
    dvalid = xgb.DMatrix(va_x[features], label=va_y)
    params = {'objective': 'binary:logistic', 'silent': 1, 'random_state': 71}
    num_round = 10  # 実際にはもっと多いround數が必要
    early_stopping_rounds = 3
    watchlist = [(dtrain, 'train'), (dvalid, 'eval')]
    model = xgb.train(params, dtrain, num_round,
                      evals=watchlist, early_stopping_rounds=early_stopping_rounds,
                      verbose_eval=0)
    va_pred = model.predict(dvalid)
    score = log_loss(va_y, va_pred)

    return score


# ---------------------------------
# Greedy Forward Selection
# ---------------------------------- 

# 需要導入模塊: from sklearn import metrics [as 別名]
# 或者: from sklearn.metrics import log_loss [as 別名]
def score(params):
    # パラメータセットを指定したときに最小化すべき関數を指定する
    # モデルのパラメータ探索においては、モデルにパラメータを指定して學習・予測させた場合のスコアとする
    model = MLP(params)
    model.fit(tr_x, tr_y, va_x, va_y)
    va_pred = model.predict(va_x)
    score = log_loss(va_y, va_pred)
    print(f'params: {params}, logloss: {score:.4f}')

    # 情報を記録しておく
    history.append((params, score))

    return {'loss': score, 'status': STATUS_OK}


# hyperoptによるパラメータ探索の実行 

# 需要導入模塊: from sklearn import metrics [as 別名]
# 或者: from sklearn.metrics import log_loss [as 別名]
def score(params):
    # パラメータを與えたときに最小化する評価指標を指定する
    # 具體的には、モデルにパラメータを指定して學習・予測させた場合のスコアを返すようにする

    # max_depthの型を整數型に修正する
    params['max_depth'] = int(params['max_depth'])

    # Modelクラスを定義しているものとする
    # Modelクラスは、fitで學習し、predictで予測値の確率を出力する
    model = Model(params)
    model.fit(tr_x, tr_y, va_x, va_y)
    va_pred = model.predict(va_x)
    score = log_loss(va_y, va_pred)
    print(f'params: {params}, logloss: {score:.4f}')

    # 情報を記録しておく
    history.append((params, score))

    return {'loss': score, 'status': STATUS_OK}


# 探索するパラメータの空間を指定する 

# 需要導入模塊: from sklearn import metrics [as 別名]
# 或者: from sklearn.metrics import log_loss [as 別名]
def make_blender_cv(classifier, x, y, calibrate=False):
    skf = StratifiedKFold(y, n_folds=5, random_state=23)
    scores, predictions = [], None
    for train_index, test_index in skf:
        if calibrate:
            # Make training and calibration
            calibrated_classifier = CalibratedClassifierCV(classifier, method='isotonic', cv=get_cv(y[train_index]))
            fitted_classifier = calibrated_classifier.fit(x[train_index, :], y[train_index])
        else:
            fitted_classifier = classifier.fit(x[train_index, :], y[train_index])
        preds = fitted_classifier.predict_proba(x[test_index, :])

        # Free memory
        calibrated_classifier, fitted_classifier = None, None
        gc.collect()

        scores.append(log_loss(y[test_index], preds))
        predictions = np.append(predictions, preds, axis=0) if predictions is not None else preds
    return scores, predictions 

# 需要導入模塊: from sklearn import metrics [as 別名]
# 或者: from sklearn.metrics import log_loss [as 別名]
def eval_pred( y_true, y_pred, eval_type):
    if eval_type == 'logloss':#eval_typeはここに追加
        loss = ll( y_true, y_pred )
        print "logloss: ", loss
        return loss            
    
    elif eval_type == 'auc':
        loss = AUC( y_true, y_pred )
        print "AUC: ", loss
        return loss             
    
    elif eval_type == 'rmse':
        loss = np.sqrt(mean_squared_error(y_true, y_pred))
        print "rmse: ", loss
        return loss




######### BaseModel Class ######### 

# 需要導入模塊: from sklearn import metrics [as 別名]
# 或者: from sklearn.metrics import log_loss [as 別名]
def logloss(y, p):
    """Bounded log loss error.

    Args:
        y (numpy.array): target
        p (numpy.array): prediction

    Returns:
        bounded log loss error
    """

    p[p < EPS] = EPS
    p[p > 1 - EPS] = 1 - EPS
    return log_loss(y, p) 

# 需要導入模塊: from sklearn import metrics [as 別名]
# 或者: from sklearn.metrics import log_loss [as 別名]
def print_k_result(ys, Ep, ll, acc, name):
    acc.append(accuracy_score(ys, Ep.argmax(axis=1)))
    ll.append(log_loss(ys, Ep))
    print("{}: accuracy = {:.4g}, log-loss = {:.4g}"
          .format(name, acc[-1], ll[-1]))