摘要: Controlling selection bias, a statistical error caused by preferential sampling of data, is fundamental problem in machine learning and inference. This paper presents simple criterion for controlling bias the odds ratio, widely used measure association between variables, that connects nature with graph modeling mechanism. If contains certain paths, we show ratio cannot be expressed using data bias. Otherwise, d-separability test can determine whether recovered, when answer affirmative, output an unbiased estimand ratio. The linear time enhances power estimand.
参考文章(10)
Alice S. Whittemore, Collapsibility of Multidimensional Contingency Tables Journal of the Royal Statistical Society: Series B (Methodological). ,vol. 40, pp. 328- 340 ,(1978) , 10.1111/J.2517-6161.1978.TB01046.X
A Method of Estimating Comparative Rates from Clinical Data. Applications to Cancer of the Lung, Breast, and Cervix Journal of the National Cancer Institute. ,vol. 11, pp. 1269- 1275 ,(1951) , 10.1093/JNCI/11.6.1269
Zhi Geng, Collapsibility of Relative Risk in Contingency Tables with a Response Variable Journal of the royal statistical society series b-methodological. ,vol. 54, pp. 585- 593 ,(1992) , 10.1111/J.2517-6161.1992.TB01898.X
Thomas Richardson, Peter Spirtes, Christopher Meek, Causal inference in the presence of latent variables and selection bias uncertainty in artificial intelligence. pp. 499- 506 ,(1995)
Silvia Acid, Luis M. De Campos, An algorithm for finding minimum d-separating sets in belief networks uncertainty in artificial intelligence. pp. 3- 10 ,(1996)
S. L. Lauritzen, A. P. Dawid, B. N. Larsen, H.-G. Leimer, Independence properties of directed markov fields Networks. ,vol. 20, pp. 491- 505 ,(1990) , 10.1002/NET.3230200503
S. Geneletti, S. Richardson, N. Best, Adjusting for selection bias in retrospective, case-control studies. Biostatistics. ,vol. 10, pp. 17- 31 ,(2008) , 10.1093/BIOSTATISTICS/KXN010
Judea Pearl, Probabilistic Reasoning in Intelligent Systems: Networks of Plausible Inference ,(1988)
Vanessa Didelez, Svend Kreiner, Niels Keiding, Graphical models for inference under outcome-dependent sampling Statistical Science. ,vol. 25, pp. 368- 387 ,(2010) , 10.1214/10-STS340
Elias Bareinboim, Judea Pearl, Controlling Selection Bias in Causal Inference international conference on artificial intelligence and statistics. pp. 100- 108 ,(2012)