feat(seaborn): implement line-win-probability

[github-actions]

Implementation: line-win-probability - seaborn

Implements the seaborn version of line-win-probability.

File: plots/line-win-probability/implementations/seaborn.py

Parent Issue: #4418

---

🤖 impl-generate workflow

[claude]

AI Review - Attempt 1/3

Image Description

The plot shows a win probability line chart for an NFL-style football game. A blue line (#306998) tracks home team win probability from play 0 to 120 against a white background. The area above the 50% dashed baseline is filled with light blue (Home), and below with light salmon/red (Away). Five key scoring events are annotated with bold colored text and connecting dots: "TD Away 7-3" (red, around play 22), "TD Home 10-7" (blue, around play 35), "TD Home 20-14" (blue, around play 65), "TD Away 23-27" (red, around play 95), and "TD Home 30-27" (blue, around play 105). Quarter markers (Q1-Q4) appear as subtle dotted vertical lines with labels at the bottom. The title reads "line-win-probability · seaborn · pyplots.ai". Y-axis shows "Home Win Probability" with percentage labels (0%-100%), X-axis shows "Play Number". A legend in the upper left identifies Home (blue) and Away (red). The final score "Home 30 – Away 27" is displayed in italic gray text at the bottom right. Top and right spines are removed, with a subtle y-axis grid.

Score: 83/100

Category	Score	Max
Visual Quality	27	30
Design Excellence	13	20
Spec Compliance	15	15
Data Quality	14	15
Code Quality	10	10
Library Mastery	4	10
Total	83	100

Visual Quality (27/30)

• VQ-01: Text Legibility (7/8) — Font sizes explicitly set (title 24, labels 20, ticks 16). Annotation text at 11pt is slightly small but still readable.
• VQ-02: No Overlap (6/6) — All annotations well-positioned with offsets, no overlapping text.
• VQ-03: Element Visibility (5/6) — Line clearly visible at 2.5 width, fills at alpha 0.25 are effective. Scatter markers at s=80 could be slightly larger for emphasis.
• VQ-04: Color Accessibility (4/4) — Blue (#306998) vs salmon-red (#D4583B) provides good contrast and is colorblind-distinguishable.
• VQ-05: Layout & Canvas (3/4) — Good overall layout. The final score text positioned at y=-0.06 falls slightly outside the axis area, creating minor imbalance at the bottom.
• VQ-06: Axis Labels & Title (2/2) — "Play Number" and "Home Win Probability" are descriptive; y-axis shows percentage format.

Design Excellence (13/20)

• DE-01: Aesthetic Sophistication (5/8) — Custom two-color palette, clean typography, frameless legend, quarter markers, annotated scores — clearly above defaults but not FiveThirtyEight-level polish. Missing: team name branding, subtitle, or more sophisticated typography hierarchy.
• DE-02: Visual Refinement (4/6) — Top/right spines removed, subtle y-grid (alpha=0.15), quarter dividers are appropriately subtle. Good refinement but could push further with background color or more polished annotation styling.
• DE-03: Data Storytelling (4/6) — Clear narrative arc: close game with lead changes, dramatic late-game swing, and final score. Color fills immediately convey momentum. Annotations highlight key turning points with running scores.

Spec Compliance (15/15)

• SC-01: Plot Type (5/5) — Correct win probability line chart.
• SC-02: Required Features (4/4) — All spec features present: 0-100% y-axis, 50% reference line, team-colored fill above/below 50%, annotated scoring events, final score display, quarter/period markers.
• SC-03: Data Mapping (3/3) — X=play number, Y=win probability. Axes show full range (0-120 plays, 0%-100%).
• SC-04: Title & Legend (3/3) — Title format "line-win-probability · seaborn · pyplots.ai" correct. Legend labels "Home"/"Away" match the data.

Data Quality (14/15)

• DQ-01: Feature Coverage (5/6) — Shows lead changes, momentum swings, multiple event types (TD, FG, INT), close game with dramatic finish. Could show a wider range of probability extremes.
• DQ-02: Realistic Context (5/5) — NFL game scenario with realistic play-by-play: touchdowns, field goals, interceptions. Neutral sports topic.
• DQ-03: Appropriate Scale (4/4) — Probabilities stay in realistic range (0.02-0.98), scoring events cause appropriate probability shifts (0.07-0.22).

Code Quality (10/10)

• CQ-01: KISS Structure (3/3) — Clean imports → data → plot → style → save flow. No functions or classes.
• CQ-02: Reproducibility (2/2) — np.random.seed(42) set.
• CQ-03: Clean Imports (2/2) — All imports used: matplotlib.patches, plt, np, pd, sns.
• CQ-04: Code Elegance (2/2) — Clean, well-organized, appropriate complexity. No fake UI.
• CQ-05: Output & API (1/1) — Saves as plot.png, dpi=300, bbox_inches='tight'. No deprecated API.

Library Mastery (4/10)

• LM-01: Idiomatic Usage (3/5) — Uses sns.lineplot axes-level API correctly, but fill_between, scatter, annotate, and all styling are pure matplotlib. Seaborn is used minimally.
• LM-02: Distinctive Features (1/5) — The only seaborn call is sns.lineplot, which is essentially identical to plt.plot. No distinctive seaborn features used (e.g., sns.set_theme, sns.set_context, seaborn color palettes, statistical features).

Score Caps Applied

• None — no caps triggered.

Strengths

• Excellent spec compliance — all required features implemented (fill areas, 50% line, annotations, quarter markers, final score)
• Strong data storytelling with a realistic NFL game narrative showing lead changes and dramatic finish
• Clean, well-organized code with good reproducibility
• Good visual refinement: removed spines, subtle grid, well-positioned annotations with running scores

Weaknesses

• Minimal seaborn usage — nearly the entire plot is built with matplotlib; only sns.lineplot uses the library
• Annotation font size (11pt) could be larger for better readability at full resolution
• Design polish could be elevated further (e.g., background tint, team name labels, subtitle)

Issues Found

1. LM-01/LM-02 LOW: Seaborn is barely used — only a single sns.lineplot call. The fill_between, scatter, annotations, and all styling are pure matplotlib.
   • Fix: Use sns.set_theme() or sns.set_context("talk") for global styling, use seaborn color palette functions (sns.color_palette), leverage seaborn's built-in styling capabilities
2. DE-01 MODERATE: Design is above defaults but not publication-quality
   • Fix: Add a subtle background color via sns.set_style("whitegrid") with customization, consider a subtitle with game context, refine annotation arrows with more sophisticated styling
3. VQ-01 MINOR: Annotation text at 11pt is slightly small
   • Fix: Increase annotation fontsize to 13-14pt

AI Feedback for Next Attempt

To reach 90+: (1) Significantly increase seaborn usage — use sns.set_theme() or sns.set_context("talk") for styling, sns.color_palette() for colors, and consider if seaborn's statistical or styling features can replace any matplotlib code. (2) Bump annotation fontsize to 13-14pt. (3) Elevate design with a subtle background tint, more refined annotation styling, or a contextual subtitle. The spec compliance and data quality are already excellent — focus improvements on library mastery and design polish.

Verdict: REJECTED

[github-actions]

🔧 Repair Attempt 1/3

Applied fixes based on AI review feedback.

Status: Repair completed, re-triggering review...

---

🤖 impl-repair

[claude]

AI Review - Attempt 2/3

Image Description

The plot displays a win probability line chart for an NFL game across 120 plays. A dark blue line (#306998) traces the home team's win probability from ~50% at kickoff through various scoring events. The area above the 50% dashed baseline is filled with translucent blue (Home) and below with translucent salmon/red (#D4583B, Away). Five key events are annotated with curved arrows and bold colored text showing the scoring play and running score: "TD Away 7-3", "TD Home 10-7", "TD Home 20-14", "TD Away 23-27", and "TD Home 30-27". Dotted vertical lines mark Q1, Q2, and Q3 boundaries. The y-axis shows 0%-100% with the 50% reference line. A styled box at the bottom right displays "Final: Home 30 – Away 27". The title follows the required format. A legend in the upper left shows Home (blue) and Away (red) patches. The background is a subtle light blue-gray (#F7F9FC). An italic subtitle reads "NFL Game — Home vs Away | Lead changes and momentum shifts across 120 plays".

Score: 91/100

Category	Score	Max
Visual Quality	29	30
Design Excellence	16	20
Spec Compliance	15	15
Data Quality	15	15
Code Quality	10	10
Library Mastery	6	10
Total	91	100

Visual Quality (29/30)

• VQ-01: Text Legibility (8/8) — All font sizes explicitly set: title 24pt, axis labels 20pt, ticks 16pt, annotations 13pt, quarter labels 14pt. All perfectly readable.
• VQ-02: No Overlap (5/6) — Q4 quarter label is hidden behind or overlaps with the "Final" score box at the bottom right. Remaining annotations are well-spaced.
• VQ-03: Element Visibility (6/6) — Line width 2.8 is clearly visible, fill areas at alpha=0.2 are distinct, scatter markers with white edge highlighting at event points.
• VQ-04: Color Accessibility (4/4) — Blue (#306998) and salmon (#D4583B) provide strong contrast and are distinguishable for colorblind users (blue-orange axis).
• VQ-05: Layout & Canvas (4/4) — 16:9 aspect ratio, plot fills canvas well, balanced margins, nothing cut off.
• VQ-06: Axis Labels & Title (2/2) — "Play Number" and "Home Win Probability" are descriptive; y-axis shows percentage units.

Design Excellence (16/20)

• DE-01: Aesthetic Sophistication (6/8) — Custom background (#F7F9FC), cohesive blue/salmon palette via sns.color_palette, italic subtitle in muted gray, styled final score box with rounded corners, arrow annotations color-matched to teams. Clearly above configured defaults.
• DE-02: Visual Refinement (5/6) — Spines removed via sns.despine, subtle y-axis grid (alpha=0.2), custom background, quarter markers as subtle dotted lines (alpha=0.6), generous whitespace.
• DE-03: Data Storytelling (5/6) — Strong narrative: annotations show scoring plays with running score, fill colors create instant momentum visualization, final score box provides closure. Viewer immediately understands the back-and-forth game.

Spec Compliance (15/15)

• SC-01: Plot Type (5/5) — Correct line chart showing win probability over game progression.
• SC-02: Required Features (4/4) — All spec features present: 50% reference line, fill above/below with team colors, annotated key events, final score display, quarter markers.
• SC-03: Data Mapping (3/3) — X=Play Number (0-120), Y=Win Probability (0%-100%). Full range displayed.
• SC-04: Title & Legend (3/3) — Title "line-win-probability · seaborn · pyplots.ai" in correct format. Legend with Home/Away color patches.

Data Quality (15/15)

• DQ-01: Feature Coverage (6/6) — Multiple lead changes, momentum swings, various event types (TD, FG, INT), both teams scoring, close competitive game.
• DQ-02: Realistic Context (5/5) — NFL game scenario with plausible play-by-play win probability. Neutral sports context.
• DQ-03: Appropriate Scale (4/4) — 120 plays realistic for NFL, probability values stay within 0.02-0.98 range, scoring swings are appropriately sized.

Code Quality (10/10)

• CQ-01: KISS Structure (3/3) — Clean imports → data generation → plot → styling → save flow. No functions or classes.
• CQ-02: Reproducibility (2/2) — np.random.seed(42) set.
• CQ-03: Clean Imports (2/2) — All imports used: matplotlib.patches, plt, np, pd, sns.
• CQ-04: Code Elegance (2/2) — Appropriate complexity, clean annotation offset dict, no fake functionality.
• CQ-05: Output & API (1/1) — Saves as plot.png with dpi=300 and bbox_inches='tight'.

Library Mastery (6/10)

• LM-01: Idiomatic Usage (4/5) — Good use of seaborn axes-level API (sns.lineplot, sns.scatterplot), sns.set_theme for global styling, sns.set_context for scaling, sns.despine for cleanup. Matplotlib supplements where seaborn lacks (fill_between, annotations).
• LM-02: Distinctive Features (2/5) — Uses seaborn-specific features (set_theme, set_context, color_palette, despine) but these are styling utilities. The core visualization (fill_between, annotations, axhline) is matplotlib. Could be replicated without seaborn with minor changes.

Score Caps Applied

• None

Strengths

• Excellent data storytelling with annotated scoring events showing running score
• Polished design with custom background, cohesive palette, and styled final score box
• Complete spec compliance — all required features implemented
• Realistic NFL game data with compelling back-and-forth narrative

Weaknesses

• Q4 quarter label is obscured by the Final score box (minor overlap issue)
• Library mastery could be stronger — core visualization relies heavily on matplotlib rather than seaborn-specific plotting features

Issues Found

1. VQ-02 MINOR: Q4 label at bottom right is hidden behind the Final score box
   • Fix: Reposition Q4 label or move Final score box to avoid overlap

AI Feedback for Next Attempt

Strong implementation at 91/100. The Q4 label overlap with the Final score box is the only visual defect. Library mastery could improve by finding more seaborn-native approaches, though this plot type inherently requires matplotlib supplementation.

Verdict: APPROVED