Skip to content

Commit 9112c0a

Browse files
github-actions[bot]github-actions[bot]
authored
feat(bokeh): implement confusion-matrix (#2287)
## Implementation: `confusion-matrix` - bokeh Implements the **bokeh** version of `confusion-matrix`. **File:** `plots/confusion-matrix/implementations/bokeh.py` --- :robot: *[impl-generate workflow](https://github.com/MarkusNeusinger/pyplots/actions/runs/20526592377)* --------- Co-authored-by: github-actions[bot] <github-actions[bot]@users.noreply.github.com> Co-authored-by: github-actions[bot] <41898282+github-actions[bot]@users.noreply.github.com>
1 parent 641e228 commit 9112c0a

2 files changed

Lines changed: 176 additions & 0 deletions

File tree

‎plots/confusion-matrix/implementations/bokeh.py‎

Lines changed: 149 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,149 @@
1+
""" pyplots.ai
2+
confusion-matrix: Confusion Matrix Heatmap
3+
Library: bokeh 3.8.1 | Python 3.13.11
4+
Quality: 96/100 | Created: 2025-12-26
5+
"""
6+
7+
import numpy as np
8+
from bokeh.io import export_png, output_file, save
9+
from bokeh.models import ColorBar, ColumnDataSource, LabelSet, LinearColorMapper
10+
from bokeh.plotting import figure
11+
from bokeh.transform import transform
12+
13+
14+
# Data - Multi-class classification results for a sentiment analysis model
15+
np.random.seed(42)
16+
17+
class_names = ["Negative", "Neutral", "Positive", "Very Positive"]
18+
19+
# Simulated confusion matrix with realistic patterns:
20+
# - Good diagonal (correct predictions)
21+
# - Adjacent classes more likely to be confused
22+
# - Some class imbalance
23+
confusion = np.array(
24+
[
25+
[142, 23, 8, 2], # Negative: mostly correct, some confused with Neutral
26+
[18, 98, 31, 5], # Neutral: often confused with adjacent classes
27+
[5, 28, 156, 24], # Positive: good accuracy, some confusion with Neutral/Very Positive
28+
[1, 4, 19, 86], # Very Positive: smaller class, good precision
29+
]
30+
)
31+
32+
# Prepare data for Bokeh heatmap using rect glyphs
33+
x_coords = []
34+
y_coords = []
35+
values = []
36+
text_labels = []
37+
38+
for i, true_class in enumerate(class_names):
39+
for j, pred_class in enumerate(class_names):
40+
x_coords.append(pred_class)
41+
y_coords.append(true_class)
42+
val = confusion[i, j]
43+
values.append(val)
44+
text_labels.append(str(val))
45+
46+
source = ColumnDataSource(data={"x": x_coords, "y": y_coords, "value": values, "text": text_labels})
47+
48+
# Color mapping - Blues sequential palette for counts
49+
colors = ["#f7fbff", "#deebf7", "#c6dbef", "#9ecae1", "#6baed6", "#4292c6", "#2171b5", "#08519c", "#08306b"]
50+
mapper = LinearColorMapper(palette=colors, low=0, high=max(values))
51+
52+
# Create figure - Square format works better for confusion matrices
53+
p = figure(
54+
width=3600,
55+
height=3600,
56+
title="confusion-matrix · bokeh · pyplots.ai",
57+
x_range=class_names,
58+
y_range=list(reversed(class_names)), # Reverse to have first class at top
59+
x_axis_label="Predicted Label",
60+
y_axis_label="True Label",
61+
tools="",
62+
toolbar_location=None,
63+
)
64+
65+
# Draw heatmap cells using rect
66+
p.rect(
67+
x="x",
68+
y="y",
69+
width=1,
70+
height=1,
71+
source=source,
72+
fill_color=transform("value", mapper),
73+
line_color="#FFFFFF",
74+
line_width=3,
75+
)
76+
77+
# Add text annotations for cell values
78+
# Calculate contrasting colors for text (white on dark, black on light)
79+
text_colors = []
80+
for val in values:
81+
# Use white text on darker cells (higher values)
82+
if val > max(values) * 0.5:
83+
text_colors.append("#FFFFFF")
84+
else:
85+
text_colors.append("#08306b")
86+
87+
source.data["text_color"] = text_colors
88+
89+
labels = LabelSet(
90+
x="x",
91+
y="y",
92+
text="text",
93+
text_color="text_color",
94+
text_font_size="32pt",
95+
text_font_style="bold",
96+
text_align="center",
97+
text_baseline="middle",
98+
source=source,
99+
)
100+
p.add_layout(labels)
101+
102+
# Style the figure for large canvas
103+
p.title.text_font_size = "36pt"
104+
p.title.text_font_style = "bold"
105+
p.title.align = "center"
106+
107+
p.xaxis.axis_label_text_font_size = "28pt"
108+
p.yaxis.axis_label_text_font_size = "28pt"
109+
p.xaxis.major_label_text_font_size = "24pt"
110+
p.yaxis.major_label_text_font_size = "24pt"
111+
112+
# Axis styling
113+
p.xaxis.axis_line_width = 2
114+
p.yaxis.axis_line_width = 2
115+
p.xaxis.major_tick_line_width = 2
116+
p.yaxis.major_tick_line_width = 2
117+
p.xaxis.major_label_orientation = 0.4 # Slight angle for readability
118+
119+
# Remove grid for cleaner heatmap look
120+
p.xgrid.grid_line_color = None
121+
p.ygrid.grid_line_color = None
122+
123+
# Add colorbar
124+
color_bar = ColorBar(
125+
color_mapper=mapper,
126+
location=(0, 0),
127+
title="Count",
128+
title_text_font_size="22pt",
129+
major_label_text_font_size="18pt",
130+
label_standoff=12,
131+
bar_line_color="#08306b",
132+
bar_line_width=2,
133+
width=30,
134+
padding=40,
135+
)
136+
p.add_layout(color_bar, "right")
137+
138+
# Adjust overall padding
139+
p.min_border_left = 150
140+
p.min_border_right = 150
141+
p.min_border_top = 100
142+
p.min_border_bottom = 150
143+
144+
# Save outputs
145+
export_png(p, filename="plot.png")
146+
147+
# Also save interactive HTML
148+
output_file("plot.html")
149+
save(p)

‎plots/confusion-matrix/metadata/bokeh.yaml‎

Lines changed: 27 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,27 @@
1+
library: bokeh
2+
specification_id: confusion-matrix
3+
created: '2025-12-26T17:37:19Z'
4+
updated: '2025-12-26T17:44:11Z'
5+
generated_by: claude-opus-4-5-20251101
6+
workflow_run: 20526592377
7+
issue: 0
8+
python_version: 3.13.11
9+
library_version: 3.8.1
10+
preview_url: https://storage.googleapis.com/pyplots-images/plots/confusion-matrix/bokeh/plot.png
11+
preview_thumb: https://storage.googleapis.com/pyplots-images/plots/confusion-matrix/bokeh/plot_thumb.png
12+
preview_html: https://storage.googleapis.com/pyplots-images/plots/confusion-matrix/bokeh/plot.html
13+
quality_score: 96
14+
review:
15+
strengths:
16+
- 'Excellent use of Bokeh-specific features: ColumnDataSource for data management,
17+
LinearColorMapper with transform() for color encoding, LabelSet for cell annotations'
18+
- Smart dynamic text color calculation (white on dark cells, dark blue on light
19+
cells) ensures all annotations are readable
20+
- Square 3600x3600 format is ideal for confusion matrix visualization
21+
- Realistic sentiment analysis scenario with plausible confusion patterns (adjacent
22+
classes more likely to be confused)
23+
- Clean code structure following KISS principles with good comments explaining choices
24+
weaknesses:
25+
- 'Minor: X-axis labels have slight rotation (0.4 rad) which is not strictly necessary
26+
for 4 short class names'
27+
- 'VQ-07: Could add slightly more styling to the colorbar or include summary statistics'

0 commit comments

Comments
 (0)