MMB-by-MMB Algorithm

MMB_by_MMB.py

Main Function

MMB_by_MMB(Data, target, alpha, p, maxK, verbose = False)

Input arguments:

• Data: data observation, datatype: ndarray
• target: target variable, datatype: int
• alpha: confidence threshold
• p: number of observation data nodes, datatype: int
• maxK: the maximal degree of any variable, datatype: int

Output arguments:

• P : Set of parent nodes of the target variable, datatype: ndarray
• C : Set of children nodes of the target variable, datatype: ndarray
• dis_depth1 : Set of districts nodes of the target variable, datatype: ndarray
• un : Set of nodes connected by undirected edges with the target variable, datatype: ndarray
• ci_test : The number of conditional independence tests in the program, datetype: int

Package requirements:

• numpy
• math
• scipy
• itertools

CITATION

If you use this code, please cite the following paper:

Feng Xie, Zheng Li, Peng Wu, Yan Zeng, Chunchen Liu, and Zhi Geng. Local Causal Structure Learning in the Presence of Latent Variables. The Forty-first International Conference on Machine Learning (ICML), Vienna, Austria, 2024.

If you have problems or questions, do not hesitate to send an email to xiefeng009@gmail.com or zhengli0060@gmail.com.