Mercurial > repos > goeckslab > image_learner

comparison plotly_plots.py @ 8:85e6f4b2ad18 draft default tip

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
planemo upload for repository https://github.com/goeckslab/gleam.git commit 8a42eb9b33df7e1df5ad5153b380e20b910a05b6
author goeckslab
date Thu, 14 Aug 2025 14:53:10 +0000
parents
children
comparison
equal deleted inserted replaced
7:801a8b6973fb 8:85e6f4b2ad18
1 import json
2 from typing import Dict, List, Optional
3
4 import numpy as np
5 import plotly.graph_objects as go
6 import plotly.io as pio
7
8
9 def build_classification_plots(
10 test_stats_path: str,
11 training_stats_path: Optional[str] = None,
12 ) -> List[Dict[str, str]]:
13 """
14 Read Ludwig’s test_statistics.json and build three interactive Plotly panels:
15 - Confusion Matrix
16 - ROC-AUC
17 - Classification Report Heatmap
18
19 Returns a list of dicts, each with:
20 {
21 "title": <plot title>,
22 "html": <HTML fragment for embedding>
23 }
24 """
25 # --- Load test stats ---
26 with open(test_stats_path, "r") as f:
27 test_stats = json.load(f)
28 label_stats = test_stats["label"]
29
30 # common sizing
31 cell = 40
32 n_classes = len(label_stats["confusion_matrix"])
33 side_px = max(cell * n_classes + 200, 600)
34 common_cfg = {"displayModeBar": True, "scrollZoom": True}
35
36 plots: List[Dict[str, str]] = []
37
38 # 0) Confusion Matrix
39 cm = np.array(label_stats["confusion_matrix"], dtype=int)
40 labels = label_stats.get("labels", [str(i) for i in range(cm.shape[0])])
41 total = cm.sum()
42
43 fig_cm = go.Figure(
44 go.Heatmap(
45 z=cm,
46 x=labels,
47 y=labels,
48 colorscale="Blues",
49 showscale=True,
50 colorbar=dict(title="Count"),
51 )
52 )
53 fig_cm.update_traces(xgap=2, ygap=2)
54 fig_cm.update_layout(
55 title=dict(text="Confusion Matrix", x=0.5),
56 xaxis_title="Predicted",
57 yaxis_title="Observed",
58 yaxis_autorange="reversed",
59 width=side_px,
60 height=side_px,
61 margin=dict(t=100, l=80, r=80, b=80),
62 )
63
64 # annotate counts and percentages
65 mval = cm.max() if cm.size else 0
66 thresh = mval / 2
67 for i in range(cm.shape[0]):
68 for j in range(cm.shape[1]):
69 v = cm[i, j]
70 pct = (v / total * 100) if total > 0 else 0
71 color = "white" if v > thresh else "black"
72 fig_cm.add_annotation(
73 x=labels[j],
74 y=labels[i],
75 text=f"<b>{v}</b>",
76 showarrow=False,
77 font=dict(color=color, size=14),
78 xanchor="center",
79 yanchor="bottom",
80 yshift=2,
81 )
82 fig_cm.add_annotation(
83 x=labels[j],
84 y=labels[i],
85 text=f"{pct:.1f}%",
86 showarrow=False,
87 font=dict(color=color, size=13),
88 xanchor="center",
89 yanchor="top",
90 yshift=-2,
91 )
92
93 plots.append({
94 "title": "Confusion Matrix",
95 "html": pio.to_html(
96 fig_cm,
97 full_html=False,
98 include_plotlyjs="cdn",
99 config=common_cfg
100 )
101 })
102
103 # 2) Classification Report Heatmap
104 pcs = label_stats.get("per_class_stats", {})
105 if pcs:
106 classes = list(pcs.keys())
107 metrics = ["precision", "recall", "f1_score"]
108 z, txt = [], []
109 for c in classes:
110 row, trow = [], []
111 for m in metrics:
112 val = pcs[c].get(m, 0)
113 row.append(val)
114 trow.append(f"{val:.2f}")
115 z.append(row)
116 txt.append(trow)
117
118 fig_cr = go.Figure(
119 go.Heatmap(
120 z=z,
121 x=metrics,
122 y=[str(c) for c in classes],
123 text=txt,
124 texttemplate="%{text}",
125 colorscale="Reds",
126 showscale=True,
127 colorbar=dict(title="Value"),
128 )
129 )
130 fig_cr.update_layout(
131 title="Classification Report",
132 xaxis_title="",
133 yaxis_title="Class",
134 width=side_px,
135 height=side_px,
136 margin=dict(t=80, l=80, r=80, b=80),
137 )
138 plots.append({
139 "title": "Classification Report",
140 "html": pio.to_html(
141 fig_cr,
142 full_html=False,
143 include_plotlyjs=False,
144 config=common_cfg
145 )
146 })
147
148 return plots