Sensitivity Analysis¶

Weight sensitivity analysis classes and functions.

Classes¶

SensitivityAnalyzer¶

Main class for sensitivity analysis.

from ascicat.sensitivity import SensitivityAnalyzer

analyzer = SensitivityAnalyzer(
    calculator,        # ASCICalculator instance
    n_bootstrap=100,   # Bootstrap iterations
    random_state=42    # Reproducibility
)

Methods¶

generate_weight_grid¶

Generate systematic weight combinations.

weights = analyzer.generate_weight_grid(
    n_points=15,
    min_weight=0.1,
    max_weight=0.8
)

generate_full_simplex_grid¶

Generate complete simplex coverage.

weights = analyzer.generate_full_simplex_grid(n_points=25)

run_full_sensitivity¶

Run analysis across all weights.

results = analyzer.run_full_sensitivity(
    weights=weights,
    track_top_n=50,
    verbose=True
)

Returns dict with:

weight_results: DataFrame with per-weight metrics
rank_matrix: Catalyst ranks across weights
asci_matrix: ASCI scores across weights
symbols: Catalyst identifiers
weights: Weight combinations

compute_rank_statistics¶

Compute comprehensive rank statistics.

rank_stats = analyzer.compute_rank_statistics(results)

Returns DataFrame with:

mean_rank, std_rank, median_rank
min_rank, max_rank, rank_range
ci_lower, ci_upper, ci_width
top5_frequency, top10_frequency, top20_frequency
robustness_score

compute_sensitivity_indices¶

Compute variance-based indices.

indices = analyzer.compute_sensitivity_indices(results)
# Returns: {'S_activity': 0.35, 'S_stability': 0.30, 'S_cost': 0.35}

statistical_tests¶

Perform significance tests.

tests = analyzer.statistical_tests(results, top_n=20)
# Returns: {'friedman_test': {...}, 'kendall_w': 0.65, 'interpretation': '...'}

SensitivityVisualizer¶

Visualization for sensitivity results.

from ascicat.sensitivity import SensitivityVisualizer

viz = SensitivityVisualizer(output_dir='sensitivity/')

Methods¶

plot_ternary_heatmap¶

Weight space visualization.

viz.plot_ternary_heatmap(
    sensitivity_results,
    metric='best_asci',
    save_name='ternary'
)

plot_rank_confidence_intervals¶

Bootstrap CI plot.

viz.plot_rank_confidence_intervals(
    rank_stats,
    n_show=20,
    save_name='rank_ci'
)

plot_sensitivity_indices¶

Index bar chart.

viz.plot_sensitivity_indices(indices, save_name='indices')

plot_robustness_quadrant¶

Performance vs robustness scatter.

viz.plot_robustness_quadrant(rank_stats, save_name='quadrant')

plot_comprehensive_summary¶

Four-panel summary figure.

viz.plot_comprehensive_summary(
    sensitivity_results,
    rank_stats,
    indices,
    save_name='summary'
)

Convenience Function¶

run_enhanced_sensitivity_analysis¶

All-in-one analysis function.

from ascicat.sensitivity import run_enhanced_sensitivity_analysis

results = run_enhanced_sensitivity_analysis(
    calculator=calc,
    output_dir='sensitivity/',
    n_weight_points=15,
    n_bootstrap=100,
    verbose=True
)

Returns dict with:

sensitivity_results
rank_stats
indices
correlations
tests