HBic is a biclustering algorithm for heterogeneous and missing data.HBic handles mixed-type data, including numeric, binary, and categorical attributes. This is the source code of HBic developed in MATLAB R2020b. In addition, the Python implementation is available at py-hbic.

HBic natively handles mixed datasets with multiple mixed-data types. Some of the main characteristics of data HBic are:

• A fitness function is proposed for evaluating biclusters with mixed-type attributes and missing values.
• A model selection approach determines the most relievable biclusters based on their similarity.
• HBic automatically identifies the number of biclusters or takes this parameter as input if this knowledge is available.

HBic is described in detail in our paper:

Adán José-García, Julie Jacques, Clement Chauvet, Vincent Sobanski, and Clarisse Dhaenes  
HBIC: A Biclustering Approach for Heterogeneous Datasets
To be published in the 27TH EUROPEAN CONFERENCE ON ARTIFICIAL INTELLIGENCE.
https://www.ecai2024.eu/

Getting Started

HBic was developed with MATLAB. To try the algorithm, look at the scripts demo_heterogeneous_data.m and demo_numerical_data.m.

Contact us

Adán José-García (adan.jose-garcia@univ-lille.fr)