MP-CTR: Workflow Reconfiguration Algorithms
Description
This code implements mutual-prediction-driven constrained topology reconfiguration (MP-CTR), including two-agent behavior prediction, workflow reconfiguration, constrained decision-making, state updates, and availability evaluation. It also includes R-HEFT, NSGA-II, and MCTS as comparison algorithms.
Files
Steps to reproduce
1.Download and extract MP-CTR_data.zip, preserving the directory structure. 2.Open dataset_index.csv to identify the actual random seed, number of initial states, and number of valid records for each dataset. Folder identifiers are archive labels, not random seeds. 3.Load the CSV files using Python with pandas or another CSV-compatible tool. Use intervention_dataset_method_summary.csv to examine the results of individual scenarios and compare terminal availability, mission effectiveness, and constraint violations across methods. 4.Recalculate method-level summary statistics and compare them with the supplied summary files. For paired statistical analyses, match methods by base_id and scenario, and account for interventions sharing the same initial state. 5.Use the files in 5_1_statistical_validation, 5_2_prediction_noise_robustness, and 5_3_external_baseline_comparison to examine statistical significance, prediction-noise sensitivity, and baseline comparisons. Repeat the analysis for all five datasets to assess consistency across random seeds.