Spaces:
Running
Running
Sadjad Alikhani
commited on
Update app.py
Browse files
app.py
CHANGED
@@ -62,7 +62,6 @@ def compute_f1_score(cm):
|
|
62 |
f1 = np.nan_to_num(f1) # Replace NaN with 0
|
63 |
return np.mean(f1) # Return the mean F1-score across all classes
|
64 |
|
65 |
-
# Function to plot and save confusion matrix with F1-score in the title
|
66 |
def plot_confusion_matrix_beamPred(cm, classes, title, save_path):
|
67 |
# Compute the average F1-score
|
68 |
avg_f1 = compute_f1_score(cm)
|
@@ -70,20 +69,22 @@ def plot_confusion_matrix_beamPred(cm, classes, title, save_path):
|
|
70 |
# Update title to include average F1-score
|
71 |
full_title = f"{title} (Avg F1-Score: {avg_f1:.2f})"
|
72 |
|
73 |
-
# Plot the confusion matrix
|
74 |
plt.figure(figsize=(8, 6))
|
75 |
-
plt.imshow(cm, interpolation='nearest', cmap='coolwarm')
|
76 |
-
plt.title(full_title)
|
77 |
plt.colorbar()
|
78 |
|
79 |
tick_marks = np.arange(len(classes))
|
80 |
-
plt.xticks(tick_marks, classes, rotation=45)
|
81 |
-
plt.yticks(tick_marks, classes)
|
82 |
|
83 |
-
plt.tight_layout()
|
84 |
-
plt.ylabel('True label')
|
85 |
-
plt.xlabel('Predicted label')
|
86 |
-
|
|
|
|
|
87 |
plt.close()
|
88 |
|
89 |
def compute_average_confusion_matrix(folder):
|
@@ -135,7 +136,6 @@ percentage_values_los = np.linspace(0.001, 1, 20) * 100 # 20 percentage values
|
|
135 |
from sklearn.metrics import f1_score
|
136 |
import seaborn as sns
|
137 |
|
138 |
-
# Function to compute confusion matrix, F1-score and plot it with dark mode style
|
139 |
def plot_confusion_matrix_from_csv(csv_file_path, title, save_path):
|
140 |
# Load CSV file
|
141 |
data = pd.read_csv(csv_file_path)
|
@@ -147,29 +147,30 @@ def plot_confusion_matrix_from_csv(csv_file_path, title, save_path):
|
|
147 |
# Compute confusion matrix
|
148 |
cm = confusion_matrix(y_true, y_pred)
|
149 |
|
150 |
-
#
|
151 |
-
f1 = f1_score(y_true, y_pred, average='
|
152 |
-
|
153 |
-
#
|
154 |
-
plt.style.use('dark_background')
|
155 |
plt.figure(figsize=(5, 5))
|
|
|
|
|
|
|
|
|
|
|
156 |
|
157 |
-
#
|
158 |
-
|
159 |
-
|
160 |
-
|
161 |
-
|
162 |
-
|
163 |
-
# Customize tick labels for dark mode
|
164 |
-
plt.xticks([0.5, 1.5], labels=['Class 0', 'Class 1'], color="white", fontsize=10)
|
165 |
-
plt.yticks([0.5, 1.5], labels=['Class 0', 'Class 1'], color="white", fontsize=10)
|
166 |
|
167 |
-
plt.ylabel('True label', color=
|
168 |
-
plt.xlabel('Predicted label', color=
|
169 |
-
plt.tight_layout()
|
170 |
|
171 |
# Save the plot as an image
|
172 |
-
plt.savefig(save_path,
|
173 |
plt.close()
|
174 |
|
175 |
# Return the saved image
|
@@ -304,32 +305,39 @@ def classify_based_on_distance(train_data, train_labels, test_data):
|
|
304 |
|
305 |
return torch.tensor(predictions) # Return predictions as a PyTorch tensor
|
306 |
|
307 |
-
# Function to generate confusion matrix plot
|
308 |
def plot_confusion_matrix(y_true, y_pred, title):
|
309 |
cm = confusion_matrix(y_true, y_pred)
|
|
|
|
|
|
|
|
|
|
|
310 |
plt.figure(figsize=(5, 5))
|
311 |
-
plt.imshow(cm, cmap='
|
312 |
-
plt.title(title)
|
313 |
-
plt.xlabel('Predicted')
|
314 |
-
plt.ylabel('Actual')
|
315 |
plt.colorbar()
|
316 |
-
|
317 |
-
|
318 |
-
|
319 |
-
plt.yticks([0, 1], labels=[0, 1])
|
320 |
-
|
321 |
# Annotate the confusion matrix
|
322 |
-
thresh = cm.max() / 2
|
323 |
for i in range(cm.shape[0]):
|
324 |
for j in range(cm.shape[1]):
|
325 |
plt.text(j, i, format(cm[i, j], 'd'),
|
326 |
ha="center", va="center",
|
327 |
color="white" if cm[i, j] > thresh else "black")
|
328 |
|
329 |
-
plt.
|
330 |
-
plt.
|
|
|
|
|
|
|
|
|
|
|
|
|
331 |
return Image.open(f"{title}.png")
|
332 |
|
|
|
333 |
def identical_train_test_split(output_emb, output_raw, labels, percentage):
|
334 |
N = output_emb.shape[0] # Get the total number of samples
|
335 |
|
|
|
62 |
f1 = np.nan_to_num(f1) # Replace NaN with 0
|
63 |
return np.mean(f1) # Return the mean F1-score across all classes
|
64 |
|
|
|
65 |
def plot_confusion_matrix_beamPred(cm, classes, title, save_path):
|
66 |
# Compute the average F1-score
|
67 |
avg_f1 = compute_f1_score(cm)
|
|
|
69 |
# Update title to include average F1-score
|
70 |
full_title = f"{title} (Avg F1-Score: {avg_f1:.2f})"
|
71 |
|
72 |
+
# Plot the confusion matrix with dark mode adjustments
|
73 |
plt.figure(figsize=(8, 6))
|
74 |
+
plt.imshow(cm, interpolation='nearest', cmap='coolwarm') # Dark mode color scheme
|
75 |
+
plt.title(full_title, color='white', pad=20) # Add padding to prevent title clipping, white text for dark mode
|
76 |
plt.colorbar()
|
77 |
|
78 |
tick_marks = np.arange(len(classes))
|
79 |
+
plt.xticks(tick_marks, classes, rotation=45, color='white') # White text for dark mode
|
80 |
+
plt.yticks(tick_marks, classes, color='white') # White text for dark mode
|
81 |
|
82 |
+
plt.tight_layout(pad=2.0) # Add padding to prevent axis label clipping
|
83 |
+
plt.ylabel('True label', color='white') # White text for dark mode
|
84 |
+
plt.xlabel('Predicted label', color='white') # White text for dark mode
|
85 |
+
|
86 |
+
# Save the plot with a black background for dark mode
|
87 |
+
plt.savefig(save_path, facecolor='black')
|
88 |
plt.close()
|
89 |
|
90 |
def compute_average_confusion_matrix(folder):
|
|
|
136 |
from sklearn.metrics import f1_score
|
137 |
import seaborn as sns
|
138 |
|
|
|
139 |
def plot_confusion_matrix_from_csv(csv_file_path, title, save_path):
|
140 |
# Load CSV file
|
141 |
data = pd.read_csv(csv_file_path)
|
|
|
147 |
# Compute confusion matrix
|
148 |
cm = confusion_matrix(y_true, y_pred)
|
149 |
|
150 |
+
# Calculate F1 Score
|
151 |
+
f1 = f1_score(y_true, y_pred, average='weighted')
|
152 |
+
|
153 |
+
# Plot the confusion matrix with dark mode colors
|
|
|
154 |
plt.figure(figsize=(5, 5))
|
155 |
+
plt.imshow(cm, interpolation='nearest', cmap='coolwarm') # Dark mode color scheme
|
156 |
+
plt.title(f"{title}\nF1-Score: {f1:.2f}", color='white', pad=20) # Display F1-Score in title, add padding for better visibility
|
157 |
+
plt.colorbar()
|
158 |
+
plt.xticks([0, 1], labels=['Class 0', 'Class 1'], color='white') # White text for dark mode
|
159 |
+
plt.yticks([0, 1], labels=['Class 0', 'Class 1'], color='white') # White text for dark mode
|
160 |
|
161 |
+
# Annotate the confusion matrix
|
162 |
+
thresh = cm.max() / 2
|
163 |
+
for i in range(cm.shape[0]):
|
164 |
+
for j in range(cm.shape[1]):
|
165 |
+
plt.text(j, i, format(cm[i, j], 'd'), ha="center", va="center",
|
166 |
+
color="white" if cm[i, j] > thresh else "black")
|
|
|
|
|
|
|
167 |
|
168 |
+
plt.ylabel('True label', color='white') # White text for dark mode
|
169 |
+
plt.xlabel('Predicted label', color='white') # White text for dark mode
|
170 |
+
plt.tight_layout(pad=2.0) # Add padding to prevent clipping of labels
|
171 |
|
172 |
# Save the plot as an image
|
173 |
+
plt.savefig(save_path, facecolor='black') # Set background to black for dark mode
|
174 |
plt.close()
|
175 |
|
176 |
# Return the saved image
|
|
|
305 |
|
306 |
return torch.tensor(predictions) # Return predictions as a PyTorch tensor
|
307 |
|
|
|
308 |
def plot_confusion_matrix(y_true, y_pred, title):
|
309 |
cm = confusion_matrix(y_true, y_pred)
|
310 |
+
|
311 |
+
# Calculate F1 Score
|
312 |
+
f1 = f1_score(y_true, y_pred, average='weighted')
|
313 |
+
|
314 |
+
# Plot the confusion matrix with dark mode colors
|
315 |
plt.figure(figsize=(5, 5))
|
316 |
+
plt.imshow(cm, interpolation='nearest', cmap='coolwarm') # Dark mode color scheme
|
317 |
+
plt.title(f"{title}\nF1-Score: {f1:.2f}", color='white', pad=20) # Add padding for the title to prevent clipping
|
|
|
|
|
318 |
plt.colorbar()
|
319 |
+
plt.xticks([0, 1], labels=['Class 0', 'Class 1'], color='white') # White text for dark mode
|
320 |
+
plt.yticks([0, 1], labels=['Class 0', 'Class 1'], color='white') # White text for dark mode
|
321 |
+
|
|
|
|
|
322 |
# Annotate the confusion matrix
|
323 |
+
thresh = cm.max() / 2
|
324 |
for i in range(cm.shape[0]):
|
325 |
for j in range(cm.shape[1]):
|
326 |
plt.text(j, i, format(cm[i, j], 'd'),
|
327 |
ha="center", va="center",
|
328 |
color="white" if cm[i, j] > thresh else "black")
|
329 |
|
330 |
+
plt.ylabel('True label', color='white') # White text for dark mode
|
331 |
+
plt.xlabel('Predicted label', color='white') # White text for dark mode
|
332 |
+
plt.tight_layout(pad=2.0) # Add padding to prevent clipping
|
333 |
+
|
334 |
+
# Save the plot
|
335 |
+
plt.savefig(f"{title}.png", facecolor='black') # Set background to black for dark mode
|
336 |
+
plt.close()
|
337 |
+
|
338 |
return Image.open(f"{title}.png")
|
339 |
|
340 |
+
|
341 |
def identical_train_test_split(output_emb, output_raw, labels, percentage):
|
342 |
N = output_emb.shape[0] # Get the total number of samples
|
343 |
|