An open-source package of causal feature selection and causal (Bayesian network) structure learning (Python version)
The pyCausalFS library provides access to a wide range of well-established and state-of-the-art causality-based feature selection approaches. The library is designed to facilitate the development of new algorithms in this research area and make it easier to compare new methods and existing ones available. The pyCausalFS library implements 30 representative causality-based feature selection methods. Specifically, it consists of 25 methods using conditional independence tests (16 single MB learning algorithms, 3 multiple MB learning algorithms, and 6 PC learning algorithms), and 5 score-based approaches.
1） Constraint-based MB learning methods: GSMB, IAMB, IAMBnPC, Inter-IAMB, Fast-IAMB, Inter-IAMBnPC, LRH, BAMB, FBEDk, MMMB, PCMB, HITON-MB, Semi-HITON-MB, IPCMB, STMB, MBOR
2） Multiple MB learning methods: KIAMB, TIE*(TIE and TIE_p)
3） Constraint-based PC learning methods: PC-simple, MBtoPC, HITON-PC, Semi-HITON-PC, GetPC, MMPC
4） score-based MB learning methods: SLL, S^2TMB, S^2TMB_p
5） score-based PC learning methods: SLL-PC, S^2TMB-PC Furthermore, using the pyCausalFS library, users can easily generate different local structure learning methods and local-to-global structure learning methods, which includes 3 local BN structure learning algorithms and one local-to global BN learning algorithm.
6） local BN structure learning algorithms: PCD-by-PCD, MB-by-MB, CMB
7） local-to global BN learning algorithm: MMHC
All implementation details please read the manual documentation.
References for citation:
-Kui Yu, Xianjie Guo, Lin Liu, Jiuyong Li, Hao Wang, Zhaolong Ling, and Xindong Wu. Causality-based Feature Selection: Methods and Evaluations. ACM Computing Surveys (CSUR) 53, no. 5 (2020): 1-36.
-Kui Yu, Lin Liu, and Jiuyong Li. A unified View of Causal and Non-causal Feature Selection. ACM Transactions on Knowledge Discovery from Data, in press (2020)