-
Notifications
You must be signed in to change notification settings - Fork 1
/
circular_plotter.py
554 lines (458 loc) · 23.1 KB
/
circular_plotter.py
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
import numpy as np
import pandas as pd
import matplotlib.pyplot as plt
import matplotlib.gridspec as gridspec
from sklearn.preprocessing import MinMaxScaler
from sklearn.discriminant_analysis import LinearDiscriminantAnalysis
from sklearn.metrics import confusion_matrix
import tkinter as tk
from tkinter import filedialog
import seaborn as sns
import argparse
# Function to parse command-line arguments
def parse_args():
parser = argparse.ArgumentParser(description="Circular Plotter")
parser.add_argument("--file_path", required=True, help="Path to the CSV file")
return parser.parse_args()
class DraggableNDPoint:
def __init__(self, scc_instance):
self.scc_instance = scc_instance
self.selected_nd_point_idx = None
self.press = None
self.initial_point = None
def connect(self):
self.cidpress = self.scc_instance.ax.figure.canvas.mpl_connect(
'button_press_event', self.on_press)
self.cidrelease = self.scc_instance.ax.figure.canvas.mpl_connect(
'button_release_event', self.on_release)
self.cidmotion = self.scc_instance.ax.figure.canvas.mpl_connect(
'motion_notify_event', self.on_motion)
def on_press(self, event):
for idx, (_, _, start_point, _, row_idx, _) in enumerate(self.scc_instance.lines):
if self.scc_instance.lines[idx][0].contains(event)[0]:
self.press = True
self.selected_nd_point_idx = row_idx
self.initial_point = (event.xdata, event.ydata)
break
def on_motion(self, event):
if not self.press or self.selected_nd_point_idx is None:
return
# Calculate the new angle based on the dragged position
angle = np.arctan2(event.ydata, event.xdata)
# Compute the circle radius based on the attribute count
r = len(self.scc_instance.data.columns) - 1
r = r / (2 * np.pi)
# Move the points of the selected nD point along the circular axis
for _, _, start_point, end_point, row_idx, _ in self.scc_instance.lines:
if row_idx == self.selected_nd_point_idx:
# Update positions based on the new angle and the computed radius
start_point[0] = r * np.cos(angle)
start_point[1] = r * np.sin(angle)
end_point[0] = r * np.cos(angle + (np.pi / len(self.scc_instance.data.columns)))
end_point[1] = r * np.sin(angle + (np.pi / len(self.scc_instance.data.columns)))
# Refresh the SCCWithChords plot
self.scc_instance.update_plot()
def on_release(self, event):
self.press = None
self.selected_nd_point_idx = None
self.initial_point = None
class DraggableLine:
def __init__(self, line):
self.line = line
self.press = None
self.background = None
def connect(self):
'connect to all the events we need'
self.cidpress = self.line.figure.canvas.mpl_connect(
'button_press_event', self.on_press)
self.cidrelease = self.line.figure.canvas.mpl_connect(
'button_release_event', self.on_release)
self.cidmotion = self.line.figure.canvas.mpl_connect(
'motion_notify_event', self.on_motion)
def update_predictions(self, angle):
scc_instance = self.line.axes.figure.scc_instance # assuming scc_instance is attached to figure
X = scc_instance.data.drop(columns='class').values
y = scc_instance.data['class'].values
y_pred = []
for i, point in enumerate(X):
side = self.line.axes.figure.scc_instance.point_side_of_line(point, angle)
if side == 'left':
y_pred.append(0) # assuming 0 represents one class
else:
y_pred.append(1) # assuming 1 represents the other class
cm = confusion_matrix(y, y_pred)
accuracy = np.mean(y == y_pred)
accuracy_title = f"Confusion Matrix\nAccuracy: {accuracy:.2%}"
ax2 = self.line.axes.figure.axes[1] # assuming ax2 is the second axes in the figure
ax2.clear()
sns.heatmap(cm, annot=True, fmt="d", cmap="Blues", ax=ax2, cbar=False)
ax2.set_xlabel('Predicted labels')
ax2.set_ylabel('True labels')
ax2.set_title(accuracy_title)
self.line.figure.canvas.draw()
def on_press(self, event):
'on button press we will see if the mouse is over us and store some data'
if event.inaxes != self.line.axes: return
contains, attrd = self.line.contains(event)
if not contains: return
self.press = True
def on_motion(self, event):
'on motion we will move the line if the mouse is over us'
if self.press is None: return
if event.inaxes != self.line.axes: return
# Only change the angle of rotation, not the position
x0, y0 = 0, 0
dx = event.xdata - x0
dy = event.ydata - y0
angle = np.arctan2(dy, dx)
r = 2 # fixed distance
x = r * np.cos(angle)
y = r * np.sin(angle)
self.line.set_data([x0, x], [y0, y])
self.update_predictions(angle)
self.line.figure.canvas.draw()
def on_release(self, event):
'on release we reset the press data'
self.press = None
self.line.figure.canvas.draw()
def disconnect(self):
'disconnect all the stored connection ids'
self.line.figure.canvas.mpl_disconnect(self.cidpress)
self.line.figure.canvas.mpl_disconnect(self.cidrelease)
self.line.figure.canvas.mpl_disconnect(self.cidmotion)
def lighten_color(color, factor=0.2):
"""
Lightens the given color.
Parameters:
color : tuple of float
Original color as an (R, G, B) tuple.
factor : float
Factor to lighten the color.
0 means no change, 1 means white color.
Default is 0.2.
Returns:
tuple of float
Lightened color as an (R, G, B) tuple.
"""
r = color[0] + (1 - color[0]) * factor
g = color[1] + (1 - color[1]) * factor
b = color[2] + (1 - color[2]) * factor
return (r, g, b)
def adjusted_bezier_curve(p0, p1, class_order, radius, coef=100):
"""Calculate quadratic Bezier curve points with control points adjusted based on attribute count."""
# Calculate the midpoint between p0 and p1
x = (p0[0] + p1[0]) / 2
y = (p0[1] + p1[1]) / 2
# Calculate the distance between the center point (0,0) and the midpoint
mDistance = np.sqrt(np.power(-x, 2) + np.power(-y, 2))
if class_order == 0: # Inner curve
# Calculate the scaling factor for the inner curve
factor = 0.1 * coef / 100
else: # Outer curve
# Calculate the scaling factor for the outer curve
factor = 4 * coef / 100
mScale = factor * radius / mDistance
# Calculate the control point
control_point = np.array([x * mScale, y * mScale])
# Invert the direction for the second class (or any subsequent class)
if class_order > 0:
direction = -control_point / np.linalg.norm(control_point)
control_point = control_point + 2 * direction * mDistance
# Calculate the Bezier curve points
t_values = np.linspace(0, 1, 10)
curve_points = []
for t in t_values:
point = (1 - t) ** 2 * p0 + 2 * (1 - t) * t * control_point + t ** 2 * p1
curve_points.append(point)
return np.array(curve_points)
class SCCWithChords:
def __init__(self, dataframe):
self.data = dataframe
self.positions = []
self.colors = []
self.selected_point_idx = None # Initialize here
self.selected_class_idx = None # Initialize here
self.transform_data()
def transform_data(self):
# The transformation code remains the same as the previous class
attribute_count = self.data.shape[1] - 1
scaler = MinMaxScaler((0, 1))
class_column_index = self.data.columns.get_loc('class')
numeric_data = self.data.drop(columns='class')
class_data = self.data.iloc[:, class_column_index]
numeric_data = scaler.fit_transform(numeric_data)
self.data.iloc[:, [i for i in range(self.data.shape[1]) if i != class_column_index]] = numeric_data
section_array = np.linspace(0, 1, attribute_count)
classes = self.data['class'].unique()
color_palette = sns.color_palette('husl', len(classes))
self.color_map = dict(zip(classes, color_palette))
self.all_positions = []
self.all_colors = []
for class_order, class_name in enumerate(classes):
class_positions = []
class_colors = []
df_name = self.data[self.data['class'] == class_name]
for index, row in df_name.iterrows():
positions = []
colors = []
y_values = row.drop('class').values
x_coord = np.linspace(0, 1, attribute_count)
arc_length = 0
for i, y in enumerate(y_values):
arc_length += y
radius = attribute_count / (2 * np.pi)
center_angle = arc_length * 360 / (2 * np.pi * radius)
center_angle = np.pi * center_angle / 180
x = radius * np.sin(center_angle)
y = radius * np.cos(center_angle)
positions.append([x, y])
colors.append(self.color_map[class_name])
class_positions.extend(positions)
class_colors.extend(colors)
self.all_positions.append(class_positions)
self.all_colors.append(class_colors)
def on_press(self, event):
"""Called when the mouse button is pressed."""
# Check if the cursor is on top of a point
for idx, (positions, colors, class_name) in enumerate(zip(self.all_positions, self.all_colors, self.data['class'].unique())):
distances = np.linalg.norm(np.array(positions) - [event.xdata, event.ydata], axis=1)
if np.min(distances) < 0.1: # 0.1 is a threshold to determine if a point is under the cursor
self.selected_point_idx = np.argmin(distances)
self.selected_class_idx = idx
break
def on_release(self, event):
"""Called when the mouse button is released."""
self.selected_point_idx = None
self.selected_class_idx = None
def on_drag(self, event):
"""Called when the mouse is dragged."""
if self.selected_point_idx is None or self.selected_class_idx is None:
return
# Compute the new position on the circle
angle = np.arctan2(event.ydata, event.xdata)
r = len(self.data.columns) - 1 # the radius
x = r * np.cos(angle)
y = r * np.sin(angle)
# Update the point's position in the plot
self.all_positions[self.selected_class_idx][self.selected_point_idx] = [x, y]
self.update_plot()
# Update the dataset (This step is a bit more involved because you need to compute the new attribute value based on the new position)
def update_plot(self):
# Recreate your plot based on the updated data points
# For simplicity, let's just clear the existing plot and redraw.
# You can optimize this if needed.
self.ax.clear()
for class_order, (positions, colors) in enumerate(zip(self.all_positions, self.all_colors)):
positions = np.array(positions)
self.ax.scatter(positions[:, 0], positions[:, 1], color=colors, s=20, alpha=0.5)
lightened_color = lighten_color(colors[0])
for i in range(0, len(positions) - 1, self.attribute_count):
for j in range(i, i + self.attribute_count - 1): # Connect each point to the next within the row
start_pos = positions[j]
end_pos = positions[j + 1]
curve_points = adjusted_bezier_curve(start_pos, end_pos, class_order, self.circle_radius)
line, = self.ax.plot(curve_points[:, 0], curve_points[:, 1], color=lightened_color, alpha=0.3)
self.lines.append((line, lightened_color, start_pos, end_pos, j // self.attribute_count, class_order))
# Additional code to plot other elements like circles, labels, etc.
self.ax.figure.canvas.draw()
def point_side_of_line(self, point, angle):
print("Debug: ", point)
#x, y = point
#coef = [np.cos(angle), np.sin(angle)]
#boundary = -coef[0] / coef[1]
# y = mx is the line equation; if y > mx, then point is above the line (or to the right)
#return 'right' if y > boundary * x else 'left'
def update_predictions(self, angle):
X = self.line.axes.figure.scc_instance.data.drop(columns='class').values
y = self.line.axes.figure.scc_instance.data['class'].values
# Dictionaries to store counts
left_counts = {}
right_counts = {}
total_counts = {}
# Determine side of each data point
for i, point in enumerate(X):
side = self.line.axes.figure.scc_instance.point_side_of_line(point, angle)
class_name = y[i]
if side == 'left':
left_counts[class_name] = left_counts.get(class_name, 0) + 1
else:
right_counts[class_name] = right_counts.get(class_name, 0) + 1
total_counts[class_name] = total_counts.get(class_name, 0) + 1
# Compute percentages
left_percentages = {k: (v / total_counts[k]) * 100 for k, v in left_counts.items()}
right_percentages = {k: (v / total_counts[k]) * 100 for k, v in right_counts.items()}
# Display the results
print("Counts & Percentages:")
for class_name in total_counts:
print(f"Class {class_name}:")
print(f"Left of Boundary: {left_counts.get(class_name, 0)} ({left_percentages.get(class_name, 0):.2f}%)")
print(f"Right of Boundary: {right_counts.get(class_name, 0)} ({right_percentages.get(class_name, 0):.2f}%)")
print("-" * 40)
cm = confusion_matrix(y, self.y_pred)
accuracy = np.mean(y == self.y_pred)
accuracy_title = f"Confusion Matrix\nAccuracy: {accuracy:.2%}"
ax2 = self.line.axes.figure.axes[1] # assuming ax2 is the second axes in the figure
ax2.clear()
sns.heatmap(cm, annot=True, fmt="d", cmap="Blues", ax=ax2, cbar=False)
ax2.set_xlabel('Predicted labels')
ax2.set_ylabel('True labels')
ax2.set_title(accuracy_title)
def on_hover(self, event):
"""Called when the mouse moves over the figure."""
info_texts = [] # List to store the hover details for all highlighted lines
hovered_row = None # Keep track of which data row is being hovered over
hovered_class = None # Keep track of which class the hovered data point belongs to
for line, original_color, start_point, end_point, row, class_info in self.lines:
if line.contains(event)[0]:
hovered_row = row
hovered_class = class_info
break
for line, original_color, start_point, end_point, row, class_info in self.lines:
if row == hovered_row and class_info == hovered_class:
line.set_color('yellow')
line.set_alpha(1.0)
line.set_zorder(1) # Bring the line to the front
else:
line.set_color(original_color)
line.set_alpha(0.4)
line.set_zorder(0)
# Retrieve the correct vector information
if hovered_row is not None:
class_names = self.data['class'].unique()
class_name = class_names[hovered_class]
vector = self.data[self.data['class'] == class_name].iloc[hovered_row].values
info_texts.append(str(vector))
# Fix the hover info box position to the top-left corner of the axes
self.hover_info_box.set_position((0, 1))
self.hover_info_box.set_ha('left')
self.hover_info_box.set_va('top')
self.hover_info_box.set_transform(self.ax.transAxes) # Using the axes coordinates
# Update the textbox with the vector representation
self.hover_info_box.set_text("\n\n".join(info_texts))
plt.draw()
def plot(self, lda=None, dataset_name=None):
fig = plt.figure(figsize=(12, 8)) # Adjusted the figure size for better layout
fig.scc_instance = self
# Connect mouse events to the methods
fig.canvas.mpl_connect('button_press_event', self.on_press)
fig.canvas.mpl_connect('button_release_event', self.on_release)
fig.canvas.mpl_connect('motion_notify_event', self.on_drag)
# Define gridspec to create a grid layout
gs = gridspec.GridSpec(1, 2, width_ratios=[4, 1]) # Adjusted to have the confusion matrix on the right
ax = plt.subplot(gs[0]) # The main visualization will be on the left
ax2 = plt.subplot(gs[1]) # The confusion matrix will be on the right
self.ax = ax
self.attribute_count = self.data.shape[1] - 1
self.circle_radius = self.attribute_count / (2 * np.pi)
self.lines = [] # To store the plotted lines for hover effect
# Close the plot on 'Escape' or 'Ctrl+W' key press
def close_on_key(event):
if event.key == 'escape' or event.key == 'ctrl+w':
plt.close(fig)
fig.canvas.mpl_connect('key_press_event', close_on_key)
if dataset_name:
ax.set_title(f"{dataset_name} in Dynamic Circular Coordinates")
for class_order, (positions, colors) in enumerate(zip(self.all_positions, self.all_colors)):
positions = np.array(positions)
ax.scatter(positions[:, 0], positions[:, 1], color=colors, s=20, alpha=0.5)
lightened_color = lighten_color(colors[0])
for i in range(0, len(positions) - 1, self.attribute_count):
for j in range(i, i + self.attribute_count - 1): # Connect each point to the next within the row
start_pos = positions[j]
end_pos = positions[j + 1]
curve_points = adjusted_bezier_curve(start_pos, end_pos, class_order, self.circle_radius)
line, = ax.plot(curve_points[:, 0], curve_points[:, 1], color=lightened_color, alpha=0.3)
self.lines.append((line, lightened_color, start_pos, end_pos, j // self.attribute_count, class_order))
# Connect the motion_notify_event to the on_hover function
fig.canvas.mpl_connect('motion_notify_event', self.on_hover)
circle = plt.Circle((0, 0), self.circle_radius, color='darkgrey', fill=False)
ax.add_artist(circle)
# Adjust the label positioning and color
attributes = self.data.drop(columns='class').columns
for i, attribute in enumerate(attributes):
angle = 2 * np.pi * (i + 0.5) / self.attribute_count
label_radius = (self.circle_radius + self.attribute_count / 4)
x = label_radius * np.sin(angle)
y = label_radius * np.cos(angle)
if 0 <= angle <= np.pi:
ha = 'left'
else:
ha = 'right'
ax.text(x, y, attribute, ha=ha, va='center', rotation=0, color='darkgrey')
# Style changes: dark grey coordinate axes and labels
ax.spines['left'].set_color('darkgrey')
ax.spines['left'].set_linewidth(0.5)
ax.spines['bottom'].set_color('darkgrey')
ax.spines['bottom'].set_linewidth(0.5)
ax.spines['right'].set_color('none')
ax.spines['top'].set_color('none')
ax.yaxis.tick_left()
ax.xaxis.tick_bottom()
ax.xaxis.set_tick_params(width=0.5)
ax.yaxis.set_tick_params(width=0.5)
plt.xticks(color='darkgrey')
plt.yticks(color='darkgrey')
# Calculate the LDA discrimination line's angle
coef = lda.coef_[0]
m = -coef[0] / coef[1]
theta = np.arctan(m)
self.circle_radius *= 2
# Draw the LDA discrimination line as a radial line
x_end = self.circle_radius * np.cos(theta)
y_end = self.circle_radius * np.sin(theta)
# Replace the LDA discrimination line plotting code with this:
lda_line, = self.ax.plot([0, x_end], [0, y_end], color='black', linestyle='--', picker=5) # 5 points tolerance
draggable = DraggableLine(lda_line)
draggable.connect()
# Label the LDA boundary position
boundary_label_position_factor = 1.1 # adjust this factor to place the label slightly outside the circle
x_label = boundary_label_position_factor * x_end
y_label = boundary_label_position_factor * y_end
self.ax.text(x_label, y_label, "LDA Boundary", fontsize=8, ha='center')
# Plot the confusion matrix in ax2
X = self.data.drop(columns='class').values
y = self.data['class'].values
self.y_pred = lda.predict(X)
accuracy = np.mean(y == self.y_pred)
cm = confusion_matrix(y, self.y_pred)
accuracy_title = f"Confusion Matrix\nAccuracy: {accuracy:.2%}"
sns.heatmap(cm, annot=True, fmt="d", cmap="Blues", ax=ax2, cbar=False)
ax2.set_xlabel('Predicted labels')
ax2.set_ylabel('True labels')
ax2.set_title(accuracy_title)
# Create a textbox for hover details
self.hover_info_box = ax.text(0.0, 0.0, '', transform=ax.transAxes, fontsize=8,
bbox=dict(facecolor='whitesmoke', edgecolor='darkgrey', alpha=0.2, boxstyle='round'))
# Add legend for class color notation
for class_name, color in self.color_map.items():
ax.plot([], [], ' ', label=class_name, marker='o', color=color, markersize=10, markeredgecolor="none")
ax.legend(loc="best", frameon=False, title="Classes")
ax.set_xlim([-0.25-self.attribute_count / 4, 0.25+self.attribute_count / 4])
ax.set_ylim([-0.25-self.attribute_count / 4, 0.25+self.attribute_count / 4])
ax.set_aspect('equal')
plt.show()
def load_and_visualize():
args = parse_args()
file_path = args.file_path
dataset_name = args.file_path.split('/')[-1].split('.')[0]
# Load data from the provided file_path
df = pd.read_csv(file_path)
scc_instance = SCCWithChords(df)
# Fit LDA and predict
X = df.drop(columns='class').values
y = df['class'].values
lda = LinearDiscriminantAnalysis()
lda.fit(X, y)
# After fitting the LDA and predicting:
y_pred = lda.predict(X)
misclassified = np.where(y != y_pred)[0]
# decision_boundary = None
# if len(misclassified) > 0:
# misclassified_positions = [point[0] for idx in misclassified for point in SCCWithChords.all_positions[idx]]
# leftmost = min(misclassified_positions, key=lambda x: x[0]) # Assuming x is the x-coordinate
# rightmost = max(misclassified_positions, key=lambda x: x[0]) # Assuming x is the x-coordinate
# # Calculate the decision boundary
# decision_boundary = (leftmost[0] + rightmost[0]) / 2
scc_instance.plot(lda, dataset_name=dataset_name)
if __name__ == "__main__":
load_and_visualize()