Python metrics.confusion_matrix方法代码示例

# 需要导入模块: from sklearn import metrics [as 别名]
# 或者: from sklearn.metrics import confusion_matrix [as 别名]
def plot_confusion_matrix(y_true, y_pred, size=None, normalize=False):
    """plot_confusion_matrix."""
    cm = confusion_matrix(y_true, y_pred)
    fmt = "%d"
    if normalize:
        cm = cm.astype('float') / cm.sum(axis=1)[:, np.newaxis]
        fmt = "%.2f"
    xticklabels = list(sorted(set(y_pred)))
    yticklabels = list(sorted(set(y_true)))
    if size is not None:
        plt.figure(figsize=(size, size))
    heatmap(cm, xlabel='Predicted label', ylabel='True label',
            xticklabels=xticklabels, yticklabels=yticklabels,
            cmap=plt.cm.Blues, fmt=fmt)
    if normalize:
        plt.title("Confusion matrix (norm.)")
    else:
        plt.title("Confusion matrix")
    plt.gca().invert_yaxis() 

# 需要导入模块: from sklearn import metrics [as 别名]
# 或者: from sklearn.metrics import confusion_matrix [as 别名]
def class_accuracy(prediction, label):
    cf = confusion_matrix(prediction, label)
    cls_cnt = cf.sum(axis=1)
    cls_hit = np.diag(cf)

    cls_acc = cls_hit / cls_cnt.astype(float)

    mean_cls_acc = cls_acc.mean()

    return cls_acc, mean_cls_acc 

# 需要导入模块: from sklearn import metrics [as 别名]
# 或者: from sklearn.metrics import confusion_matrix [as 别名]
def train_and_evaluate(clf, X_train, X_test, y_train, y_test):
    clf.fit(X_train, y_train)
    print ("Accuracy on training set:")
    print (clf.score(X_train, y_train))
    print ("Accuracy on testing set:")
    print (clf.score(X_test, y_test))
    y_pred = clf.predict(X_test)
    print ("Classification Report:")
    print (metrics.classification_report(y_test, y_pred))
    print ("Confusion Matrix:")
    print (metrics.confusion_matrix(y_test, y_pred))


# ===============================================================================
# from FaceDetectPredict.py
# =============================================================================== 

# 需要导入模块: from sklearn import metrics [as 别名]
# 或者: from sklearn.metrics import confusion_matrix [as 别名]
def draw_confusion_matrix(dataset, model, set_trial=None, filename="test_results.sdf"):
    path = find_average_trial(dataset, model, metric="test_pr") if set_trial is None \
        else "../result/{}/{}/{}/".format(model, dataset, set_trial)

    # Load true, pred value
    true_y, pred_y = [], []
    mols = Chem.SDMolSupplier(path + filename)

    for mol in mols:
        true_y.append(float(mol.GetProp("true")))
        pred_y.append(float(mol.GetProp("pred")))

    true_y = np.array(true_y, dtype=float)
    pred_y = np.array(pred_y, dtype=float).round()

    # Get precision and recall
    confusion = confusion_matrix(true_y, pred_y)
    tn, fp, fn, tp = confusion.ravel()

    print("tn: {}, fp: {}, fn: {}, tp: {}".format(tn, fp, fn, tp)) 

# 需要导入模块: from sklearn import metrics [as 别名]
# 或者: from sklearn.metrics import confusion_matrix [as 别名]
def classification_scores(gts, preds, labels):
    accuracy        = metrics.accuracy_score(gts,  preds)
    class_accuracies = []
    for lab in labels: # TODO Fix
        class_accuracies.append(metrics.accuracy_score(gts[gts == lab], preds[gts == lab]))
    class_accuracies = np.array(class_accuracies)

    f1_micro        = metrics.f1_score(gts,        preds, average='micro')
    precision_micro = metrics.precision_score(gts, preds, average='micro')
    recall_micro    = metrics.recall_score(gts,    preds, average='micro')
    f1_macro        = metrics.f1_score(gts,        preds, average='macro')
    precision_macro = metrics.precision_score(gts, preds, average='macro')
    recall_macro    = metrics.recall_score(gts,    preds, average='macro')

    # class wise score
    f1s        = metrics.f1_score(gts,        preds, average=None)
    precisions = metrics.precision_score(gts, preds, average=None)
    recalls    = metrics.recall_score(gts,    preds, average=None)

    confusion = metrics.confusion_matrix(gts,preds, labels=labels)

    #TODO confusion matrix, recall, precision
    return accuracy, f1_micro, precision_micro, recall_micro, f1_macro, precision_macro, recall_macro, confusion, class_accuracies, f1s, precisions, recalls 

# 需要导入模块: from sklearn import metrics [as 别名]
# 或者: from sklearn.metrics import confusion_matrix [as 别名]
def main(argv):
    parser = argparse.ArgumentParser()
    parser.add_argument("--classifier-saved-model-path", type=str)
    parser.add_argument("--text-file-path", type=str, required=True)
    parser.add_argument("--label-index", type=str, required=False)
    parser.add_argument("--label-file-path", type=str, required=False)
    args_namespace = parser.parse_args(argv)
    command_line_args = vars(args_namespace)

    global logger
    logger = log_initializer.setup_custom_logger(global_config.logger_name, "INFO")

    if not command_line_args['label_file_path'] and not command_line_args['label_index']:
        raise Exception("Provide either label-index or label_file_path")

    [style_transfer_score, confusion_matrix] = \
        get_style_transfer_score(command_line_args['classifier_saved_model_path'],
                                 command_line_args['text_file_path'],
                                 command_line_args['label_index'], 
                                 command_line_args['label_file_path'])
    logger.info("style_transfer_score: {}".format(style_transfer_score))
    logger.info("confusion_matrix: {}".format(confusion_matrix)) 

# 需要导入模块: from sklearn import metrics [as 别名]
# 或者: from sklearn.metrics import confusion_matrix [as 别名]
def plot_confusion(title, true_labels, predicted_labels, normalized=True):
    labels = list(set(true_labels) | set(predicted_labels))

    if normalized:
        cm = confusion_matrix(true_labels, predicted_labels, labels=labels)
        cm = cm.astype('float') / cm.sum(axis=1)[:, np.newaxis]
    else:
        cm = confusion_matrix(true_labels, predicted_labels, labels=labels)

    fig, ax = plt.subplots(figsize=(10, 10))
    ax.imshow(cm, interpolation='nearest', cmap=plt.cm.Blues)
    ax.set_title(title)
    # plt.colorbar()
    tick_marks = np.arange(len(labels))
    ax.set_xticks(tick_marks)
    ax.set_xticklabels(labels, rotation=90)
    ax.set_yticks(tick_marks)
    ax.set_yticklabels(labels)
    ax.set_ylabel('True Label')
    ax.set_xlabel('Predicted Label')
    ax.grid(False)
    return fig, ax 

# 需要导入模块: from sklearn import metrics [as 别名]
# 或者: from sklearn.metrics import confusion_matrix [as 别名]
def QuadWeightedKappa(y, y_pred):
  y_pred = np.argmax(y_pred, 1)
  cm = confusion_matrix(y, y_pred)
  classes_y, counts_y = np.unique(y, return_counts=True)
  classes_y_pred, counts_y_pred = np.unique(y_pred, return_counts=True)
  E = np.zeros((classes_y.shape[0], classes_y.shape[0]))
  for i, c1 in enumerate(classes_y):
    for j, c2 in enumerate(classes_y_pred):
      E[c1, c2] = counts_y[i] * counts_y_pred[j]
  E = E / np.sum(E) * np.sum(cm)
  w = np.zeros((classes_y.shape[0], classes_y.shape[0]))
  for i in range(classes_y.shape[0]):
    for j in range(classes_y.shape[0]):
      w[i, j] = float((i - j)**2) / (classes_y.shape[0] - 1)**2
  re = 1 - np.sum(w * cm) / np.sum(w * E)
  return re 

# 需要导入模块: from sklearn import metrics [as 别名]
# 或者: from sklearn.metrics import confusion_matrix [as 别名]
def _report_ice_cloud(self, output_dir, experiment, test, retrieved):
        # Confusion matrix:
        fig, ax = plt.subplots(figsize=(12, 10))
        cm = confusion_matrix(test.ice_cloud, retrieved.ice_cloud)
        img = self._plot_matrix(cm, classes=["Yes", "No"], normalize=True)
        fig.colorbar(img, label="probability")
        ax.set_title("Ice Cloud Classifier - Performance")
        ax.set_ylabel('real ice cloud')
        ax.set_xlabel('predicted ice cloud')
        fig.tight_layout()
        fig.savefig(join(output_dir, "ice-cloud-confusion-matrix.png"))

        fig, ax = plt.subplots(figsize=(12, 10))
        ax.barh(
            np.arange(len(self.ice_cloud.inputs)),
            self.ice_cloud.estimator.feature_importances_
        )
        ax.set_yticks(np.arange(len(self.ice_cloud.inputs)))
        ax.set_yticklabels(self.ice_cloud.inputs)
        ax.set_xlabel("Feature Importance")
        ax.set_ylabel("Feature")
        ax.set_title("Ice Cloud Classifier - Importance")
        fig.savefig(join(output_dir, "ice-cloud-feature-importance.png")) 

# 需要导入模块: from sklearn import metrics [as 别名]
# 或者: from sklearn.metrics import confusion_matrix [as 别名]
def find_optimal(error_df):
    optimal_threshold = 1000
    max_f1 = 0
    max_pr = 0
    max_re = 0
    for threshold in range(1000, 400000, 5000):
        print("Threshold: " + str(threshold))
        y_pred = [1 if e > threshold else 0 for e in error_df.Reconstruction_error.values]
        conf_matrix = confusion_matrix(error_df.True_class, y_pred)
        precision, recall, f1 = compute_metrics(conf_matrix)
        if f1 > max_f1:
            max_f1 = f1
            optimal_threshold = threshold
            max_pr = precision
            max_re = recall
    return optimal_threshold, max_pr, max_re, max_f1 

# 需要导入模块: from sklearn import metrics [as 别名]
# 或者: from sklearn.metrics import confusion_matrix [as 别名]
def print_evaluation(model,data,ls,log=None):
    features,actual = data
    predictions = predict(model, features, 500).data.numpy().reshape(-1).tolist()

    labels = [ls.idx[i] for i, _ in enumerate(ls.idx)]

    actual = [labels[i] for i in actual]
    predictions = [labels[i] for i in predictions]

    print(accuracy_score(actual, predictions))
    print(classification_report(actual, predictions))
    print(confusion_matrix(actual, predictions))

    data = zip(actual,predictions)
    if log is not None:
        f = open(log, "w+")
        for a,p in data:
            f.write(json.dumps({"actual": a, "predicted": p}) + "\n")
        f.close() 

# 需要导入模块: from sklearn import metrics [as 别名]
# 或者: from sklearn.metrics import confusion_matrix [as 别名]
def run_story_evaluation(story_file, policy_model_path, nlu_model_path,
                         out_file, max_stories):
    """Run the evaluation of the stories, plots the results."""
    from sklearn.metrics import confusion_matrix
    from sklearn.utils.multiclass import unique_labels

    test_y, preds = collect_story_predictions(story_file, policy_model_path,
                                              nlu_model_path, max_stories)

    log_evaluation_table(test_y, preds)
    cnf_matrix = confusion_matrix(test_y, preds)
    plot_confusion_matrix(cnf_matrix, classes=unique_labels(test_y, preds),
                          title='Action Confusion matrix')

    fig = plt.gcf()
    fig.set_size_inches(int(20), int(20))
    fig.savefig(out_file, bbox_inches='tight') 

# 需要导入模块: from sklearn import metrics [as 别名]
# 或者: from sklearn.metrics import confusion_matrix [as 别名]
def evaluate(config, model, data_iter, test=False):
    model.eval()
    loss_total = 0
    predict_all = np.array([], dtype=int)
    labels_all = np.array([], dtype=int)
    with torch.no_grad():
        for texts, labels in data_iter:
            outputs = model(texts)
            loss = F.cross_entropy(outputs, labels)
            loss_total += loss
            labels = labels.data.cpu().numpy()
            predic = torch.max(outputs.data, 1)[1].cpu().numpy()
            labels_all = np.append(labels_all, labels)
            predict_all = np.append(predict_all, predic)

    acc = metrics.accuracy_score(labels_all, predict_all)
    if test:
        report = metrics.classification_report(labels_all, predict_all, target_names=config.class_list, digits=4)
        confusion = metrics.confusion_matrix(labels_all, predict_all)
        return acc, loss_total / len(data_iter), report, confusion
    return acc, loss_total / len(data_iter) 

# 需要导入模块: from sklearn import metrics [as 别名]
# 或者: from sklearn.metrics import confusion_matrix [as 别名]
def accuracy(y_true, y_pred):        
    # 计算混淆矩阵
    y = np.zeros(len(y_true))
    y_ = np.zeros(len(y_true))    
    for i in range(len(y_true)): 
        y[i] = np.argmax(y_true[i,:])
        y_[i] = np.argmax(y_pred[i,:])
    cnf_mat = confusion_matrix(y, y_)
    
    # Acc = 1.0*(cnf_mat[1][1]+cnf_mat[0][0])/len(y_true)
    # Sens = 1.0*cnf_mat[1][1]/(cnf_mat[1][1]+cnf_mat[1][0])
    # Spec = 1.0*cnf_mat[0][0]/(cnf_mat[0][0]+cnf_mat[0][1])
    
    # # 绘制ROC曲线
    # fpr, tpr, thresholds = roc_curve(y_true[:,0], y_pred[:,0])
    # Auc = auc(fpr, tpr)
    
    
    # 计算多分类评价值
    Sens = recall_score(y, y_, average='macro')
    Prec = precision_score(y, y_, average='macro')
    F1 = f1_score(y, y_, average='weighted') 
    Support = precision_recall_fscore_support(y, y_, beta=0.5, average=None)
    return Sens, Prec, F1, cnf_mat 

# 需要导入模块: from sklearn import metrics [as 别名]
# 或者: from sklearn.metrics import confusion_matrix [as 别名]
def save_cnf_roc(y_true, y_pred, classes, isPlot, save_tag = ''):
    # 计算混淆矩阵
    y = np.zeros(len(y_true))
    y_ = np.zeros(len(y_true))    
    for i in range(len(y_true)): 
        y[i] = np.argmax(y_true[i,:])
        y_[i] = np.argmax(y_pred[i,:])
    cnf_mat = confusion_matrix(y, y_)
    print cnf_mat
    
    # # 记录混淆矩阵
    f = open('experiments/img/confuse_matrixes.txt', 'ab+')
    if save_tag[-1] == '0':
        f.write(save_tag+'\n')
    f.write('No.' + save_tag[-1] + '\n')
    f.write(str(cnf_mat) + '\n')
    f.close()

    # # 记录ROC曲线
    plot_roc_curve(y_true, y_pred, range(classes), 'all/'+save_tag)  

###########################
# 计算TP、TN、FP、FN