Replication data for Hyperedge prediction and the statistical mechanisms of higher-order and lower-order interactions in complex networks

This dataset contains all the input data and results necessary to reproduce the experiments described in the associated publication on hyperedge prediction and statistical mechanisms in complex networks.

### Folder structure

#### `data/DATA_FOLDS/`

This directory contains the data used for 5-fold cross-validation of the model. Each fold consists of a pair of files:

- `trainX.dat`: the training data for fold X (where X ∈ [0–4])
- `testX.dat`: the corresponding test data for the same fold

Each training set represents 80% of the original dataset, and each test set represents the remaining 20%. The data files are in `.dat` format and contain encoded hyperedges used to fit and evaluate the generative model.

#### `data/REDUCED_DEFINITIVE_RESULTS/`

This directory includes the prediction results for each of the five folds (`fold0` to `fold4`) for different values of the model parameter **K**, which determines the number of latent groups:

- `K2_foldX.csv` → prediction results for fold X using K = 2
- `K3_foldX.csv` → results using K = 3
- ...
- `K5_foldX.csv` → results using K = 5

All result files are in `.csv` format and contain metrics and predicted scores generated by the model.

#### `README.md`

This file describes the usage of the dataset and details the experimental setup.

### Notes

The dataset is provided in support of reproducibility and open research. The full version of this dataset is also structured in a public GitHub repository:
https://github.com/AleixMT/TrigenicInteractionPredictor

All data is released under a Creative Commons Attribution 4.0 International License (CC BY 4.0).