Academic
Publications
Scalable Techniques for Mining Causal Structures

Scalable Techniques for Mining Causal Structures,Data Mining and Knowledge Discovery,Craig Silverstein,Sergey Brin,Rajeev Motwani,Jeffrey D. Ullman

Scalable Techniques for Mining Causal Structures   (Citations: 34)
BibTex | RIS | RefWorks Download
Journal: Data Mining and Knowledge Discovery - DATAMINE , vol. 4, no. 2/3, pp. 163-192, 2000
Cumulative Annual
    • ...Algorithms for learning association rules evaluate whether new items are unexpectedly correlated with a target item conditioned on the existing items in the rule [3]...

    Matthew J. H. Rattiganet al. Leveraging D-Separation for Relational Data Sets

    • ...“In our view, inferring complete causal models (i.e., causal Bayesian networks) is essentially impossible in large-scale data mining applications with thousands of variables” (Silverstein et al., 2000)...

    Ioannis Tsamardinoset al. The max-min hill-climbing Bayesian network structure learning algorith...

    • ...Recent developments in causal inference or causal statistics makes the assignment of cause and effect possible, if the third variab le is available and information on conditional correlation can be obtained (Cooper, 1997; Spirtes et al., 2000; Pearl, 2000; Silverstein et al., 2000)...
    • ...The LCD rule is extended in (Silverstein et al., 2000) by requiring that x and z are correlated unconditional on y (but uncorrelated conditional on anti-CCP RF SE No. samples + + + 960 + + - 128 + - + 84 + - - 19 - + + 95 - + - 74 - - + 214 - - - 149...
    • ...Note that the CCC rule in (Silverstein et al., 2000) that does not allow for hidden variables is not discussed here, since the assumption for it s use is violated in our example...
    • ...This is the so-called CCU rule discussed in (Silverstein et al., 2000)...
    • ...We have successfully applied a local causality discovery rule (Cooper, 1997; Silverstein et al., 2000) to the three-variable set of two biomarkers for rheumatoid arthritis, anti-cyclic ci trullinated peptide antibody and rheumatoid factor, and one genotype known to be associated with the rheumatoid arthritis, the HLA-DRB1 allele...

    Wentian Liet al. Inferring causal relationships among intermediate phenotypes and bioma...

    • ...A Local Causal Discovery (LCD [11]) algorithm ([8]) is used to study how causal structures can be determined from association rules and generate rules to map symptoms to diseases...

    Shibendra S. Pobiet al. Predicting Juvenile Diabetes from Clinical Test Results

    • ...Following the framework of [7], we assume that our data was generated by a causal directed acyclic graph, where an edge A → B has the meaning that “A is a direct cause of B”. There are several advantages on trying to extract subgraphs of the original graph as a type of association rule, instead of discovering a full graph [7], as further discussed in Section 2.1...
    • ...Following the framework of [7], we assume that our data was generated by a causal directed acyclic graph, where an edge A → B has the meaning that “A is a direct cause of B”. There are several advantages on trying to extract subgraphs of the original graph as a type of association rule, instead of discovering a full graph [7], as further discussed in Section 2.1...
    • ...What do we gain by extracting association rules from a graphical model instead of trying to learn the graphical structure directly? One major reason is scalability, as motivated by [7]: the data might have been generated by a directed graph that is too large to be efficiently learned from data...

    Ricardo Silvaet al. Towards Association Rules with Hidden Variables

Sort by: