English | Español

---

All figures are generated by run_analysis.py --phase figures (and outputs for the outputs heatmap). They are stored in analysis/figures/ (PNG and PDF). Links below point to the repo; if the wiki is published from a clone, you can upload these images to the wiki or use raw URLs.

Base path in repo: analysis/figures/
Raw URL pattern: https://raw.githubusercontent.com/Radega1993/the-one-scenario-corpus/main/analysis/figures/<filename>.png

---

Correlation matrix between scenario feature vectors (60x60, corpus_v1). Each cell is the Pearson r between two scenarios. Red = high positive correlation; blue = negative. High off-diagonal values indicate pairs that are linearly similar in parameter space.

File: heatmap_pearson.png

---

Same as above but Spearman (rank) correlation. Useful to check robustness to non-linear monotonic relationships.

File: heatmap_spearman.png

---

Distribution of pairwise Pearson |r| across all scenario pairs (1770 pairs, 60 scenarios). Shows how many pairs fall in each correlation bin. Goal: most mass below 0.7.

File: histogram_correlations_pearson.png

---

Distribution of pairwise Spearman |r|. Compare with Pearson to see if rank correlation is similarly spread.

File: histogram_correlations_spearman.png

---

First two principal components of the normalised feature matrix Z. Each point is a scenario. Shows how scenarios are distributed in feature space; clusters and outliers are visible.

File: scatter_pca_regression.png

---

The two scenarios that have the highest Pearson |r| (most correlated pair). One scenario’s normalised features on the x-axis, the other on the y-axis. Points near the diagonal indicate strong linear agreement.

File: scatter_max_r_pair_regression.png

---

Correlation matrix between scenario output vectors (delivery ratio, latency mean, overhead ratio, drop ratio from ONE). Shows whether scenario behaviour (not only parameters) is redundant.

File: heatmap_pearson_outputs.png

---

File	Description
correlation_report.txt	Max |r|, mean |r|, pairs with |r| ≥ 0.7, Spearman summary, cosine/euclidean distances, diversity criteria, top correlated pairs, FDR/Bonferroni.
multiple_comparisons_report.txt	Rejections and pairs with high |r| significant after FDR and Bonferroni.
clustering_report.txt	Ward clustering summary (e.g. k=7), silhouette.
outputs_correlation_report.txt	Correlation and distances on output vectors.
scenarios_to_diversify.txt	List of scenarios to diversify (high correlation or dense clusters).
observaciones_correlacion.md	Notes on correlation and benchmark criteria.
plan_radical_scenarios.md	Plan for radical scenarios (spatial structure, TTL, buffer, etc.).
cambios_diversificacion_top_pares.md	Record of diversification changes for top pairs.

Links: reports/ in repo

---

Main CSV files produced by the pipeline:

File	Description
features.csv	One row per scenario, columns = features (46).
features_normalized.csv	Z-score normalised features.
normalization_params.csv	Mean and std per feature.
correlation_pearson.csv, correlation_spearman.csv	60x60 correlation matrices (feature vectors).
distance_cosine.csv, distance_euclidean.csv	60x60 distance matrices.
cluster_assignments.csv	Scenario → cluster (Ward).
output_metrics.csv	Delivery ratio, latency mean, overhead ratio, drop ratio per scenario.
correlation_pearson_outputs.csv, distance_*_outputs.csv	Correlation/distance on output vectors.

Links: data/ in repo

---

• Results overview — Main metrics and interpretation
• Methodology — How figures are produced
• Quickstart — Regenerate with --phase figures and --phase outputs