Hyper-Geometric Networks

Hyper-Geometric Networks is an Association Rule Mining algorithm suitable for datasets with mixed data types(both continuous and discrete variables) and unknown functions. This algorithm uses Hyper-Geometric distribution to statistical test and find frequent datasets.

Note

Unlike most Association Rule Mining algorithms, Hyper-Geometric Networks algorithm is not a Rule-Based System but an Statistical Model.


Advantages:

  • The results are highly explainable and interpretable.
  • Works with mixed data types, so it can be applied to various domains.
  • It's robust to noisy data and sparse distributions, and can handle missing values.

Disadvantages:

  • It can be computationally heavy with larger datasets.

References: