Replication Data and Source Code for: Comparative Evaluation of Best-Fit Nesting Heuristics for FDM 3D Printing Batch Scheduling

Preprocessing pipeline (run in order): step1_filter.py Geometric filter: applies five sequential criteria (face count, watertightness, planar surface, minimum size, build volume compatibility) to raw Thingi10K STL files step2_reposition.py Flat reposition: aligns parts with non-printable default orientations to the build platform using Rodrigues' rotation formula step3_orientation.py — Orientation optimization: selects optimal build orientation for each part using a scoring function balancing base contact area and build height Nesting algorithms: naive_method.py Naive sequential row-fill baseline (no sorting, no placement optimization) naive_sorted.py Sorted Naive baseline (size-sorted row-fill; enables sorting decomposition) best_fit.py Best-Fit spatial criterion (BF) best_fit_height_aware.py Height-penalized Best-Fit (BF-H) best_fit_grouped.py Volume-grouped Best-Fit (BF-G) Data files: nesting_input.csv Processed dataset of 5,335 geometrically valid, orientation-optimized parts derived from Thingi10K experiment_results_v3.csv Results of 25 paired experimental runs (batch count, processing time, platform utilization, mean maximum height per batch) for all five algorithms. statistical_results.xlsx Full statistical analysis output including Shapiro-Wilk normality tests, Wilcoxon signed-rank tests (vs. Naive, sorting decomposition, placement decomposition, and pairwise BF variants), and Repeated Measures ANOVA results Experiment runner and analysis: design_experiment.py Main experiment runner: draws 25 paired random samples from nesting_input.csv and executes all five algorithms on each sample, recording results to experiment_results_v3.csv statistical_analysis.py Statistical analysis script: computes Shapiro-Wilk normality tests, Wilcoxon signed-rank tests (all comparison levels), Repeated Measures ANOVA, and pairwise post-hoc comparisons; outputs statistical_results.xlsx