FEI Nina, YANG Youlong†School of Mathematics and Statistics, Xidian University, Xi’an 710126, Shaanxi, China
Finding causality merely from observed data is a fundamental problem in science. The most basic form of this causal problem is to determine whether X leads to Y or Y leads to X in the case of joint observation of two variables X, Y. In statistics, path analysis is used to describe the direct dependence between a set of variables. But in fact, we usually do not know the causal order between variables. However, ignoring the direction of the causal path will prevent researchers from analyzing or using causal models. In this study, we propose a method for estimating causality based on observed data. First, observed variables are cleaned and valid variables are retained. Then, a direct linear non-Gaussian acyclic graph models (DirectLiNGAM) estimates the causal order K between variables. The third step is to estimate the adjacency matrix B of the causal relationship based on K. Next, since B is not convenient for model interpretation, we use adaptive lasso to prune the causal path and variables. Further, a causal path graph and a recursive model are established. Finally, we test and debug the recursive model, obtain a causal model with good fit, and estimate the direct, indirect and total effects between causal variables. This paper overcomes the randomness assigning causal order to variables. This study is different from the researcher’s understanding of his own model by generating some form of simulation data. The simplest and relatively unsmooth statistical learning method used in this study has obvious advantages in the field of interpretable machine learning.
 Wright S. The relative importance of heredity and environ-ment in determining the piebald pattern of Guinea-pigs[J]. Proceedings of the National Academy of Sciences, 1920, 6(6): 320-332.
 Wright S. Correlation and causation[J]. Journal of Agricul-tural Research, 1921, 20(7): 557-585.
 Laan V D , Mark J. Causal inference for a population of causally connected units[J]. Journal of Causal Inference, 2014, 2(1):62.
 Shimizu S, Hoyer P, Hyvärinen A, et al. A linear non-Gaussian acyclic model for causal discovery[J]. Journal of Machine Learning Research, 2006, 7(4): 2003-2030.
 Pearl J. Causality: Models, Reasoning, and Inference[M]. Cambridge : Cambridge University Press, 2000.
 Spirtes P, Glymour C N, Scheines R, et al. Causation, Pre-diction, and Search[M]. Cambridge : MIT Press, 2000.
 Zhang L Q, Cichocki A, Amari S. Natural gradient algorithm for blind separation of over determined mixture with additive noise[J]. IEEE Signal Processing Letters, 1999, 6(11): 293- 295.
 Hyvärinen A. Fast and robust fixed-point algorithms for independent component analysis[J]. IEEE Transactions on Neural Networks, 1999, 10(3): 626.
 Himberg J, Hyvärinen A, Esposito F. Validating the inde-pendent components of neuroimaging time series via clustering and visualization[J]. Neuroimage, 2004, 22(3): 1214- 1222.
 Shimizu S, Hyvärinen A, Kawahara Y, et al. A direct method for estimating a causal ordering in a linear non-Gaussian acyclic model[C]//Proceedings of the Twenty-Fifth Conference on Uncertainty in Artificial Intelligence. Montreal: AUAI Press, 2009: 506-513. arXiv:1408.2038
 Bollen K A. Structural Equations with Latent Variables[M]. New York: John Wiley & Sons, 1989: 289-308.
 Hoyer P O, Janzing D, Mooij J, et al. Nonlinear causal dis-covery with additive noise models[C]//International Con-ference on Neural Information Processing Systems. Vancouver: Curran Associates Inc, 2008: 689-696.
 Evermann J, Tate M. Assessing the predictive performance of structural equation model estimators[J]. Journal of Business Research, 2016, 69(10): 4565-4582.
 Mooij J M, Peters J, Janzing D, et al. Distinguishing cause from effect using observational data: Methods and benchmarks[J]. Journal of Machine Learning Research, 2016, 17(1):1103-1204.
 Blau P M, Duncan O D. Some preliminary findings on social stratification in the United States[J]. Acta Sociologica, 1965, 9(1/2):4-24.
 Peters J, Mooij J M, Janzing D. Causal discovery with con-tinuous additive noise models[J]. Journal of Machine Learning Research, 2014, 15(1): 2009-2053.
 Shimizu S , Inazumi T , Sogawa Y , et al. DirectLiNGAM: A direct method for learning a linear non-Gaussian structural equation model[J]. Journal of Machine Learning Research, 2011, 12(4):1225-1248.
 Zou H. The adaptive lasso and its oracle properties[J]. Jour-nal of Industrial and Management Optimization (JIMO), 2006, 101(476):1418-1429.
 Rong T S. AMOS and Research Methods[M]. Chongqing: Chongqing University Press, 2010:120-128(Ch).