Question

我正在尝试绘制三个具有更多面部识别样本（http://vis-www.cs.umass.edu/lfw/）的面部的ROC曲线。内核分析，拟合和图形均正常运行（显然）。问题是在制作ROC曲线时出现的，这看起来很奇怪，我相信不是那样，有人可以帮助我吗？

print(__doc__)

# Display progress logs on stdout
logging.basicConfig(level=logging.INFO, format='%(asctime)s %(message)s')

# #############################################################################
# Download the data, if not already on disk and load it as numpy arrays

lfw_people = fetch_lfw_people(min_faces_per_person=143, resize=0.4)

# introspect the images arrays to find the shapes (for plotting)
n_samples, h, w = lfw_people.images.shape

# for machine learning we use the 2 data directly (as relative pixel
# positions info is ignored by this model)
X = lfw_people.data

n_features = X.shape[1]

# the label to predict is the id of the person
y = lfw_people.target
target_names = lfw_people.target_names
n_classes = target_names.shape[0]

print("Total dataset size:")
print("n_samples: %d" % n_samples)
print("n_features: %d" % n_features)
print("n_classes: %d" % n_classes)

#############################################################################
# Split into a training set and a test set using a stratified k fold

# split into a training and testing set
X_train, X_test, y_train, y_test = train_test_split(
    X, y, test_size=0.25, random_state=42)

# #############################################################################
# Compute a PCA (eigenfaces) on the 01face dataset (treated as unlabeled
# dataset): unsupervised feature extraction / dimensionality reduction
n_components = 150

print("Extracting the top %d eigenfaces from %d faces"
      % (n_components, X_train.shape[0]))
t0 = time()
pca = PCA(n_components=n_components, svd_solver='randomized',
          whiten=True).fit(X_train)
print("done in %0.3fs" % (time() - t0))

eigenfaces = pca.components_.reshape((n_components, h, w))

print("Projecting the input data on the eigenfaces orthonormal basis")
t0 = time()
X_train_pca = pca.transform(X_train)
X_test_pca = pca.transform(X_test)
print("done in %0.3fs" % (time() - t0))

# #############################################################################
# Train a SVM classification model

print("Fitting the linear classifier to the training set")
t0 = time()
param_grid = {'C': [1e3, 5e3, 1e4, 5e4, 1e5],
              'gamma': [0.0001, 0.0005, 0.001, 0.005, 0.01, 0.1], }

clf_linear = GridSearchCV(SVC(kernel='linear', class_weight='balanced'),
                          param_grid, cv=5)
clf_linear = clf_linear.fit(X_train_pca, y_train)
print("done in %0.3fs" % (time() - t0))
print("Best estimator found by grid search:")
print(clf_linear.best_estimator_)

print("Fitting the rbf classifier to the training set")
t0 = time()
clf_rbf = GridSearchCV(SVC(kernel='rbf', class_weight='balanced'),
                       param_grid, cv=5)
clf_rbf = clf_rbf.fit(X_train_pca, y_train)
print("done in %0.3fs" % (time() - t0))
print("Best estimator found by grid search:")
print(clf_rbf.best_estimator_)

# #############################################################################
# Quantitative evaluation of the model quality on the test set

print("Predicting people's names on the test set (linear kernel)")
t0 = time()
y_pred_linear = clf_linear.predict(X_test_pca)
print("done in %0.3fs" % (time() - t0))

linear_confusion_matrix = confusion_matrix(y_test, y_pred_linear, labels=range(n_classes))

print(classification_report(y_test, y_pred_linear, target_names=target_names))
print("Matriz de Confusão - Kernel Linear")
print(linear_confusion_matrix)

print("Predicting people's names on the test set (rbf kernel)")
t0 = time()
y_pred_rbf = clf_rbf.predict(X_test_pca)
print("done in %0.3fs" % (time() - t0))

rbf_confusion_matrix = confusion_matrix(y_test, y_pred_rbf, labels=range(n_classes))

print(classification_report(y_test, y_pred_rbf, target_names=target_names))
print("Matriz de Confusão - Kernel RBF")
print(rbf_confusion_matrix)


# #############################################################################
# Qualitative evaluation of the predictions using matplotlib

def plot_gallery(images, titles, h, w, n_row=3, n_col=4):
    """Helper function to plot a gallery of portraits"""
    plt.figure(figsize=(1.8 * n_col, 2.4 * n_row))
    plt.subplots_adjust(bottom=0, left=.01, right=.99, top=.90, hspace=.35)
    for i in range(n_row * n_col):
        plt.subplot(n_row, n_col, i + 1)
        plt.imshow(images[i].reshape((h, w)), cmap=plt.cm.gray)
        plt.title(titles[i], size=12)
        plt.xticks(())
        plt.yticks(())


# plot the result of the prediction on a portion of the test set

def title(y_pred, y_test, target_names, i):
    pred_name = target_names[y_pred[i]].rsplit(' ', 1)[-1]
    true_name = target_names[y_test[i]].rsplit(' ', 1)[-1]
    return 'predicted: %s\ntrue:      %s' % (pred_name, true_name)


prediction_titles_linear = [title(y_pred_linear, y_test, target_names, i)
                            for i in range(y_pred_linear.shape[0])]

plot_gallery(X_test, prediction_titles_linear, h, w)

prediction_titles_rbf = [title(y_pred_rbf, y_test, target_names, i)
                         for i in range(y_pred_rbf.shape[0])]

plot_gallery(X_test, prediction_titles_rbf, h, w)

# plot the gallery of the most significative eigenfaces

eigenface_titles = ["eigenface %d" % i for i in range(eigenfaces.shape[0])]
plot_gallery(eigenfaces, eigenface_titles, h, w)

plt.show()

# Binarizing the outputs
y_test = y = label_binarize(y_test, classes=[0, 1, 2])
y_pred_linear = y = label_binarize(y_pred_linear, classes=[0, 1, 2])
y_pred_rbf = y = label_binarize(y_pred_rbf, classes=[0, 1, 2])


# Compute ROC curve and ROC area for linear kernel
fpr_linear = dict()
tpr_linear = dict()
roc_auc_linear = dict()

for i in range(n_classes):
    fpr_linear[i], tpr_linear[i], _ = roc_curve(y_test[:, i], y_pred_linear[:, i])
    roc_auc_linear[i] = auc(fpr_linear[i], tpr_linear[i])

# Compute micro-average ROC curve and ROC area for linear kernel
fpr_linear["micro"], tpr_linear["micro"], _ = roc_curve(y_test.ravel(), y_pred_linear.ravel())
roc_auc_linear["micro"] = auc(fpr_linear["micro"], tpr_linear["micro"])

# Compute ROC curve and ROC area for rbf kernel
fpr_rbf = dict()
tpr_rbf = dict()
roc_auc_rbf = dict()

for i in range(n_classes):
    fpr_rbf[i], tpr_rbf[i], _ = roc_curve(y_test[:, i], y_pred_rbf[:, i])
    roc_auc_rbf[i] = auc(fpr_rbf[i], tpr_rbf[i])

# Compute micro-average ROC curve and ROC area for rbf kernel
fpr_rbf["micro"], tpr_rbf["micro"], _ = roc_curve(y_test.ravel(), y_pred_rbf.ravel())
roc_auc_rbf["micro"] = auc(fpr_rbf["micro"], tpr_rbf["micro"])

# Plotting and comparing the ROCs curves
plt.figure()
lw = 2
plt.plot(fpr_linear[2], tpr_linear[2], color='darkorange', lw=lw,
         label='Curva ROC - Kernel Linear(area = %0.5f)' % roc_auc_linear[2])
plt.plot(fpr_rbf[2], tpr_rbf[2], color='navy', lw=lw,
         label='Curva ROC - Kernel RBF(area = %0.5f)' % roc_auc_rbf[2], linestyle='--')
plt.xlim([0.0, 1.0])
plt.ylim([0.0, 1.0])
plt.xlabel('Taxa Falso Positivo')
plt.ylabel('Taxa Verdadeiro Positivo')
plt.title('Comparação entre o kernel linear e rbf pela Curva ROC')
plt.legend(loc="lower right")
plt.show()

：中华民国身材

该ROC曲线的图形看起来很奇怪，人脸识别sklearn

0 个答案: