| """ |
| Visualize AFRES v2 results. |
| """ |
| import json |
| import numpy as np |
| import matplotlib |
| matplotlib.use('Agg') |
| import matplotlib.pyplot as plt |
|
|
| def load(path="/app/afres_v2_results.json"): |
| with open(path) as f: |
| return json.load(f) |
|
|
| def plot_search_history(results, save="/app/v2_search_history.png"): |
| hist = results.get("search_history", []) |
| if not hist: |
| return |
| iters = [h["iteration"] for h in hist] |
| fitnesses = [h.get("fitness", 0) for h in hist] |
| maes = [h.get("mean_mae", 0) for h in hist] |
| ics = [h.get("mean_ic", 0) for h in hist] |
|
|
| fig, (ax1, ax2, ax3) = plt.subplots(1, 3, figsize=(18, 5)) |
|
|
| ax1.scatter(iters, fitnesses, c='steelblue', alpha=0.6, s=30) |
| ax1.plot(sorted(set(iters)), [max(f for i, f in zip(iters, fitnesses) if i == ii) |
| for ii in sorted(set(iters))], 'r-', linewidth=2) |
| ax1.set_xlabel("Iteration") |
| ax1.set_ylabel("Fitness (Mean IC)") |
| ax1.set_title("Rubric Search: Fitness Landscape") |
| ax1.grid(True, alpha=0.3) |
|
|
| ax2.scatter(iters, ics, c='green', alpha=0.6, s=30) |
| ax2.set_xlabel("Iteration") |
| ax2.set_ylabel("Mean IC") |
| ax2.set_title("Mean IC per Evaluated Rubric") |
| ax2.grid(True, alpha=0.3) |
|
|
| ax3.scatter(iters, maes, c='orange', alpha=0.6, s=30) |
| ax3.set_xlabel("Iteration") |
| ax3.set_ylabel("Mean MAE") |
| ax3.set_title("Mean Regression MAE per Evaluated Rubric") |
| ax3.grid(True, alpha=0.3) |
|
|
| plt.tight_layout() |
| plt.savefig(save, dpi=150) |
| plt.close() |
| print(f"Saved {save}") |
|
|
| def plot_rubric_composition(results, save="/app/v2_rubric_composition.png"): |
| rubric = results.get("best_rubric", {}) |
| items = rubric.get("items", []) |
| if not items: |
| return |
|
|
| fig, ax = plt.subplots(figsize=(10, 6)) |
| labels = [i["id"] for i in items] |
| n_features = [len(i.get("feature_focus", [])) for i in items] |
| n_ops = [len(i.get("preferred_ops", [])) for i in items] |
|
|
| x = np.arange(len(labels)) |
| width = 0.35 |
| ax.bar(x - width/2, n_features, width, label='# Features', color='steelblue') |
| ax.bar(x + width/2, n_ops, width, label='# Operators', color='coral') |
| ax.set_xticks(x) |
| ax.set_xticklabels(labels, rotation=45, ha='right') |
| ax.set_ylabel("Count") |
| ax.set_title("Best Rubric: Constraint Composition") |
| ax.legend() |
| ax.grid(True, alpha=0.3, axis='y') |
|
|
| plt.tight_layout() |
| plt.savefig(save, dpi=150) |
| plt.close() |
| print(f"Saved {save}") |
|
|
| def plot_factor_ic_distribution(results, save="/app/v2_ic_distribution.png"): |
| """Plot IC distribution of all evaluated factors.""" |
| hist = results.get("search_history", []) |
| all_ics = [] |
| for h in hist: |
| |
| pass |
|
|
| |
| ics = [h.get("mean_ic", 0) for h in hist if h.get("mean_ic", 0) > 0] |
| if len(ics) < 2: |
| return |
|
|
| fig, ax = plt.subplots(figsize=(10, 5)) |
| ax.hist(ics, bins=20, color='steelblue', edgecolor='black', alpha=0.7) |
| ax.axvline(np.mean(ics), color='red', linestyle='--', linewidth=2, label=f'Mean={np.mean(ics):.4f}') |
| ax.set_xlabel("Mean IC of Rubric-Generated Factors") |
| ax.set_ylabel("Frequency") |
| ax.set_title("Distribution of Factor Quality Across Rubric Evaluations") |
| ax.legend() |
| ax.grid(True, alpha=0.3) |
|
|
| plt.tight_layout() |
| plt.savefig(save, dpi=150) |
| plt.close() |
| print(f"Saved {save}") |
|
|
| def create_report(results, save="/app/v2_report.txt"): |
| lines = [] |
| lines.append("=" * 70) |
| lines.append("AFRES v2 – Rubric-Discovery Report") |
| lines.append("=" * 70) |
| lines.append("") |
|
|
| rubric = results.get("best_rubric", {}) |
| items = rubric.get("items", []) |
| lines.append(f"BEST DISCOVERED RUBRIC ({len(items)} item(s))") |
| lines.append("-" * 50) |
| for item in items: |
| lines.append(f" [{item['id']}] {item['description']}") |
| lines.append(f" features: {item.get('feature_focus', [])}") |
| lines.append(f" operators: {item.get('preferred_ops', [])}") |
| lines.append(f" horizon: {item.get('time_horizon_hint', '?')} " |
| f"complexity: {item.get('complexity_hint', '?')}") |
| lines.append("") |
|
|
| lines.append("-" * 50) |
| lines.append(f"Best fitness (mean IC): {results.get('best_fitness', 0):.4f}") |
| lines.append(f"Total factors generated: {results.get('total_factors_generated', 0)}") |
| lines.append(f"Valid factors: {results.get('valid_factors', 0)}") |
| lines.append(f"Mean IC (all valid): {results.get('mean_ic', 0):.4f}") |
| lines.append(f"Mean MAE (all valid): {results.get('mean_mae', 0):.4f}") |
| lines.append(f"Mean Sharpe (all valid): {results.get('mean_sharpe', 0):.2f}") |
| lines.append("") |
|
|
| top = results.get("top_factor", {}) |
| lines.append("TOP FACTOR") |
| lines.append("-" * 50) |
| lines.append(f" ID: {top.get('id', '?')}") |
| lines.append(f" Expression: {top.get('expression', '?')}") |
| lines.append(f" IC: {top.get('ic', 0):+.4f}") |
| lines.append(f" MAE: {top.get('mae', 0):.4f}") |
| lines.append(f" R²: {top.get('r2', 0):.4f}") |
| lines.append(f" Sharpe: {top.get('sharpe', 0):.2f}") |
| lines.append(f" Returns: {top.get('returns', 0):.2f}") |
| lines.append("") |
|
|
| lines.append("=" * 70) |
| report = "\n".join(lines) |
| with open(save, "w") as f: |
| f.write(report) |
| print(f"Saved {save}") |
| return report |
|
|
| def main(): |
| results = load() |
| plot_search_history(results) |
| plot_rubric_composition(results) |
| plot_factor_ic_distribution(results) |
| report = create_report(results) |
| print("\n" + report) |
|
|
| if __name__ == "__main__": |
| main() |
|
|
