Costimize Kube - Experimental Results & Reproducibility Bundle: Predictive, Cost-Aware Kubernetes Autoscaling Evaluation (150-run matrix)

Experimental results and reproducibility bundle for Costimize Kube, a predictive, cost-aware autoscaling framework for Kubernetes (MSc dissertation). Contains the full 150-run evaluation matrix (5 scenarios × 5 approaches × 2 benchmark configurations × 3 repeats) plus 18 ablation runs, with per-run time-series CSVs, run details, summary tables, and all evaluation figures. Includes the complete experiment harness (runbook, run scripts, analysis and plotting scripts), workload manifests, a SHA-256 manifest of all 204 files, and an environment snapshot (git SHA f708588, Python 3.12, k3d cluster, uv-locked dependencies). Results regenerate via RUNBOOK.md → run_matrix.py → extract_run_details.py → analyse_results.py, all included.