Configuration Reference

To run HERA in Batch Mode, create a .json file (e.g., analysis_config.json). Below is the complete list of available parameters.

Tip

Use Default Parameters: It is strongly recommended to use the provided default parameters to ensure methodological consistency. If different parameters are used, they should be scientifically justified and clearly documented in your research methodology.

Minimal Configuration

{
  "userInput": {
    "folderPath": "/Users/Name/Data",
    "fileType": ".csv",
    "metric_names": ["Accuracy", "Runtime", "Memory"],
    "ranking_mode": "M1_M2_M3",
    "output_dir": "/Users/Name/Results"
  }
}

Full Configuration Example (JSON)

This example shows all possible parameters with their default values. Parameters inside system and bootstrap_* must be nested correctly as shown.

{
  "userInput": {
    "folderPath": "/Path/To/Data",
    "fileType": ".csv",
    "metric_names": ["Metric1", "Metric2", "Metric3"],
    "output_dir": "/Path/To/Results",
    "language": "en",
    "ranking_mode": "M1_M2_M3",
    "reproducible": true,
    "seed": 123,
    "num_workers": "auto",
    "create_reports": true,
    "plot_theme": "light",
    "ci_level": 0.95,
    "alphas": [0.05, 0.05, 0.05],
    "run_sensitivity_analysis": true,
    "run_power_analysis": true,
    "power_simulations": 10000,
    "min_data_completeness": 0.80,

    "system": {
      "target_memory": "auto",
      "jack_parfor_thr": 300,
      "jack_vec_limit": 150,
      "delta_mat_limit": 30000,
      "min_batch_size": 100
    },

    "manual_B_thr": 2000,
    "manual_B_ci": 10000,
    "manual_B_rank": 500,

    "bootstrap_seed_offset": 1000,

    "bootstrap_thresholds": {
      "B_start": 100,
      "B_step": 100,
      "B_end": 10000,
      "n_trials": 25,
      "convergence_tolerance": 0.01,
      "smoothing_window": 3,
      "convergence_streak_needed": 3,
      "min_steps_for_convergence_check": 1
    },

    "bootstrap_ci": {
      "B_start": 100,
      "B_step": 200,
      "B_end": 20000,
      "n_trials": 30,
      "convergence_tolerance": 0.03,
      "smoothing_window": 3,
      "convergence_streak_needed": 3,
      "min_steps_for_convergence_check": 1
    },

    "bootstrap_ranks": {
      "B_start": 50,
      "B_step": 25,
      "B_end": 2500,
      "n_trials": 15,
      "convergence_tolerance": 0.005,
      "smoothing_window": 3,
      "convergence_streak_needed": 3,
      "min_steps_for_convergence_check": 1
    },

    "selected_permutations": []
  }
}

Parameter Dictionary

Category	Parameter	Type	Default	Description
Input/Output	folderPath	string	-	Absolute path to data folder.
	fileType	string	-	.csv or .xlsx.
	metric_names	array	-	List of filenames (metrics) in hierarchical order.
	output_dir	string	-	Path to save results.
	language	string	"en"	Output language code.
Logic	ranking_mode	string	"M1_M2_M3"	Logic mode (M1, M1_M2, M1_M3A, M1_M2_M3).
	run_sensitivity_analysis	bool	true	Run ranking for all metric permutations.
	selected_permutations	array	[]	List of custom permutations to test (empty = all).
	run_power_analysis	bool	true	Run post-hoc power analysis.
	min_data_completeness	double	0.80	Min fraction of valid pairs required (0.8 = 80%).
Statistics	ci_level	double	0.95	Confidence interval level (e.g., 0.95 for 95%).
	alphas	array	[0.05, ...]	Significance level for each metric.
	power_simulations	int	10000	Number of simulations for power analysis.
System	reproducible	bool	true	Use fixed RNG seed.
	seed	int	123	RNG seed value.
	bootstrap_seed_offset	int	1000	Base random seed offset for parallel bootstrap operations. Prevents RNG substream collisions.
	num_workers	int/str	"auto"	Number of parallel workers. "auto" uses parcluster('local').NumWorkers.
	system.target_memory	int	"auto"	Target memory per chunk (MB). Automatically calculated based on available RAM, but can be manually defined via JSON config file.
	system.jack_parfor_thr	int	300	Min n to trigger parallel execution.
	system.jack_vec_limit	int	150	Max n for vectorized Jackknife calculations.
	system.delta_mat_limit	int	30000	Max n_x * n_y product (usually n^2) for matrix-based Cliff's Delta.
	system.min_batch_size	int	100	Min batch size for parallel processing.
Graphics	create_reports	bool	true	Generate PDF reports and high-res plots. If false no plots are saved.
	plot_theme	string	"light"	"light", "dark", "colourblind light", or "colourblind dark".
Bootstrap (Manual)	manual_B_thr	int	2000	Iterations for Thresholds (empty = auto).
	manual_B_ci	int	5000	Iterations for CIs (empty = auto).
	manual_B_rank	int	500	Iterations for Rank Stability (empty = auto).
Bootstrap (Auto)	bootstrap_thresholds	struct	(See Below)	Config for Threshold convergence.
	bootstrap_ci	struct	(See Below)	Config for CI convergence.
	bootstrap_ranks	struct	(See Below)	Config for Rank Stability convergence.

👉 Bootstrap Configuration (Auto-Convergence)