feat(letsplot): implement scatter-regression-polynomial

[github-actions]

Implementation: scatter-regression-polynomial - python/letsplot

Implements the python/letsplot version of scatter-regression-polynomial.

File: plots/scatter-regression-polynomial/implementations/python/letsplot.py

Parent Issue: #2028

---

🤖 impl-generate workflow

[claude]

AI Review - Attempt 1/3

Image Description

Light render (plot-light.png): The scatter plot displays a clear quadratic (diminishing returns) relationship between advertising spend (x-axis, 4–52 thousands $) and sales revenue (y-axis, 30–115 thousands $). The plot uses a warm off-white background (#FAF8F1) with 80 green scatter points (brand #009E73, alpha 0.65) clearly visible against the light surface. A semi-transparent green confidence band (alpha 0.15) wraps around a bold orange regression line (#D55E00, size 2.5), making the fitted curve visually distinct from the scatter. The title, axis labels (20pt), and tick labels (16pt) are all in dark text (INK colors) and highly readable. The R² value (0.936) and polynomial equation (y = -0.046x² + 3.789x + 21.45) are positioned in the top-right corner in readable gray text (INK_SOFT). Subtle grid lines (rgba 10% opacity) guide without competing with data. All chrome, text, and elements are readable and properly themed for light mode. Legibility: PASS ✓

Dark render (plot-dark.png): On the warm near-black background (#1A1A17), all text switches to light colors (INK #F0EFE8 for title/labels, INK_SOFT #B8B7B0 for ticks/secondary text), remaining fully readable with no dark-on-dark failures. The scatter points, regression line, and confidence band colors are identical to the light render (#009E73 green points, #D55E00 orange line, semi-transparent green band)—only the chrome (background, text, grid) flips theme-appropriately. The R² value and polynomial equation remain readable in light gray text. Subtle grid lines and all structural elements adapt perfectly. All title, axis label, and tick label text are clearly legible against the dark surface. Legibility: PASS ✓

Score: 93/100

Category	Score	Max
Visual Quality	30	30
Design Excellence	15	20
Spec Compliance	15	15
Data Quality	15	15
Code Quality	10	10
Library Mastery	8	10
Total	93	100

Visual Quality (30/30)

• VQ-01: Text Legibility (8/8) — All text explicitly sized and theme-adapted; readable at full resolution in both light and dark renders with zero accessibility issues
• VQ-02: No Overlap (6/6) — Title, annotations, axis labels, and tick labels all positioned without collision; R² and equation placed outside main data area
• VQ-03: Element Visibility (6/6) — 80 scatter points with alpha 0.65 clearly distinguishable; orange regression line (size 2.5) stands out; confidence band visible at alpha 0.15
• VQ-04: Color Accessibility (2/2) — Okabe-Ito positions 1 (#009E73) and 2 (#D55E00) ensure colorblind safety; excellent contrast on both light and dark surfaces
• VQ-05: Layout & Canvas (4/4) — Standard landscape 1600×900 (→4800×2700px); excellent proportions, nothing cut off, generous margins
• VQ-06: Axis Labels & Title (2/2) — "Advertising Spend (thousands $)" and "Sales Revenue (thousands $)" are descriptive with units; title follows spec format exactly
• VQ-07: Palette Compliance (2/2) — First series (#009E73 brand green) for points; regression line uses Okabe-Ito position 2; backgrounds correctly set (#FAF8F1 light / #1A1A17 dark); both renders theme-correct throughout

Design Excellence (15/20)

• DE-01: Aesthetic Sophistication (6/8) — Professional use of theme-adaptive tokens (PAGE_BG, INK, INK_SOFT, RULE); intentional color and alpha choices; polished execution beyond defaults
• DE-02: Visual Refinement (4/6) — theme_minimal() base with refined grid (opacity 10%, size 0.3); good whitespace; could gain refinement by removing top/right spines
• DE-03: Data Storytelling (5/6) — Clear visual hierarchy: green points → orange trend → confidence band → labels establish context. R² and equation support interpretation. Focal point (curved relationship) is unambiguous. Minor: confidence band could be explained more explicitly in annotations.

Spec Compliance (15/15)

• SC-01: Plot Type (5/5) — Correct scatter plot with quadratic polynomial regression; degree 2 as default; regression line is visually distinct (solid orange vs. green points)
• SC-02: Required Features (4/4) — R² prominently displayed; polynomial equation shown; confidence band with semi-transparent shading included; scatter point transparency (alpha 0.65) as specified
• SC-03: Data Mapping (3/3) — X/Y axes map independent/dependent variables correctly; all data visible (x: 4–52, y: 30–115); axes span full range
• SC-04: Title & Legend (3/3) — Title format exact: "scatter-regression-polynomial · letsplot · anyplot.ai"; legend correctly omitted (single series)

Data Quality (15/15)

• DQ-01: Feature Coverage (6/6) — Demonstrates scatter, polynomial curve, confidence band, R² statistic, polynomial equation—all key plot type aspects present
• DQ-02: Realistic Context (5/5) — Economics example (advertising spend vs. sales revenue) is plausible and neutral; diminishing returns pattern is realistic; polynomial model (y = -0.046x² + 3.789x + 21.45) realistically represents the relationship
• DQ-03: Appropriate Scale (4/4) — Advertising spend 4–50k$ (reasonable marketing budget); sales revenue 30–115k$ (proportional to spend); 80 points adequate for curve visualization; no overfitting

Code Quality (10/10)

• CQ-01: KISS Structure (3/3) — No functions/classes; straightforward linear flow (seed → data → polyfit → plot)
• CQ-02: Reproducibility (2/2) — np.random.seed(42) ensures same data on each run; polynomial fitting deterministic
• CQ-03: Clean Imports (2/2) — os, numpy, pandas, lets_plot all used; no unused imports
• CQ-04: Code Elegance (2/2) — No fake UI/interactivity; polyfit and poly1d used idiomatically; theme handling elegant with environment variable pattern
• CQ-05: Output & API (1/1) — Saves as plot-{THEME}.png and plot-{THEME}.html; current letsplot API; proper scale=3 for PNG export

Library Mastery (8/10)

• LM-01: Idiomatic Usage (5/5) — ggplot() + geom_*() composition; proper aes() mapping; geom_ribbon for confidence band; geom_text for annotations; theme_minimal() + theme() customization; ggsize() for figure sizing; ggsave() with scale parameter—all canonical letsplot patterns
• LM-02: Distinctive Features (3/5) — Uses geom_ribbon (layer-composition beyond basic scatter+line); HTML export leveraged; could explore more distinctive letsplot interactive features (callbacks, layer interactivity) for higher score

Score Caps Applied

• None — No caps triggered (VQ-02=6, VQ-03=6, SC-01=5, DQ-02=5 all above thresholds; DE-01=6 + DE-02=4 above minimum; CQ-04=2)

Strengths

• Specification compliance: All required features (scatter, polynomial curve, R², equation, confidence band) correctly implemented
• Theme-adaptive excellence: Both light and dark renders perfectly adapted—no dark-on-dark or light-on-light failures; chrome flips appropriately while data colors remain constant
• Text legibility: All text (title, labels, ticks, annotations) clearly readable at full 4800×2700px resolution in both themes
• Visual clarity: Scatter points, regression line, and confidence band are all distinctly visible; no overlaps or obscured elements
• Data realism: Economics example with diminishing returns is domain-appropriate and neutral; polynomial model realistically represents the relationship
• Color safety: Okabe-Ito palette (positions 1 & 2) ensures colorblind accessibility
• Code elegance: Simple, reproducible, idiomatic letsplot usage with proper theme handling

Weaknesses

• VQ-07 chrome refinement: Plot uses all spines; removing top/right spines (L-shaped frame) would increase visual refinement and align with anyplot minimalism guidelines
• LM-02 underutilization: letsplot has strong interactive/callback capabilities; confidence band and regression line could be interactive (hover tooltips showing predicted values, layer toggling) to showcase distinctive library features
• DE-02 minimal: Grid is subtle but could be refined further; consider removing minor gridlines or adjusting styling for enhanced visual polish

Issues Found

1. DE-02 MINOR: Spine configuration could be more refined
   • Fix: Add for top/right spines, keeping only left and bottom (L-shaped frame)
2. LM-02 MINOR: Interactive features not utilized
   • Fix: Add hover tooltips (if supported by letsplot) or layer interactivity to showcase distinctive capabilities

AI Feedback for Next Attempt

The implementation is exceptionally strong and meets all specification requirements with professional execution. To achieve 95+/100, focus on: (1) removing top/right spines for visual refinement (one line change in theme()), and (2) exploring letsplot-specific interactive features (hover tooltips on regression line showing predicted values, or layer toggle for confidence band). The current version is production-ready and exceeds median quality standards.

Verdict: APPROVED