Intelligent smoothing using hierarchical Bayesian models. Export

@article{graham_2008, abstract = {Hierarchical Bayesian modeling provides a flexible approach to modeling in multiparameter problems. Examples include disease mapping and spatiotemporal analysis, and multiple exposure modeling. A key feature of hierarchical Bayesian models is that prior expectations regarding model structure are embedded in a probability model that reflects uncertainty about the form of the structure that links analytical units (such as geographic areas). This results in posterior estimates that are compromises between raw data summaries and estimates that conform exactly to the prior model structure. The posterior estimates are more precise and generally have lower mean-squared error than traditional data summaries, and yet are not strictly constrained to follow a posited prior model form.}, author = {Graham, P.}, citeulike-article-id = {3404087}, citeulike-linkout-0 = {http://dx.doi.org/10.1097/EDE.0b013e31816b7859}, citeulike-linkout-1 = {http://view.ncbi.nlm.nih.gov/pubmed/18414089}, citeulike-linkout-2 = {http://www.hubmed.org/display.cgi?uids=18414089}, comment = {A useful explanatory paper which clearly explains what Bayesian smoothing is supposed to do, and under what circumstances it may be used. ``Observable data can be conceptualized as `structure plus noise' with the role of analysis being to reveal the structure by stripping away the noise.'' (p.493) ``When data are not plentiful for all analytical units, noise can be reduced by ” borrowing strength” or pooling information across analytical units. This requires a modeling framework that connects separate analytical units.'' (p.493) }, doi = {10.1097/EDE.0b013e31816b7859}, issn = {1044-3983}, journal = {Epidemiology}, keywords = {bayesian, commentary, hierarchical, model-structure, overview, smoothing}, month = {May}, number = {3}, pages = {493--495}, posted-at = {2008-10-14 12:21:20}, priority = {0}, title = {Intelligent smoothing using hierarchical Bayesian models.}, url = {http://dx.doi.org/10.1097/EDE.0b013e31816b7859}, volume = {19}, year = {2008} }

TY - JOUR ID - graham_2008 L3 - citeulike-article-id:3404087 N2 - Hierarchical Bayesian modeling provides a flexible approach to modeling in multiparameter problems. Examples include disease mapping and spatiotemporal analysis, and multiple exposure modeling. A key feature of hierarchical Bayesian models is that prior expectations regarding model structure are embedded in a probability model that reflects uncertainty about the form of the structure that links analytical units (such as geographic areas). This results in posterior estimates that are compromises between raw data summaries and estimates that conform exactly to the prior model structure. The posterior estimates are more precise and generally have lower mean-squared error than traditional data summaries, and yet are not strictly constrained to follow a posited prior model form. IS - 3 JF - Epidemiology SN - 1044-3983 EP - 495 TI - Intelligent smoothing using hierarchical Bayesian models. VL - 19 SP - 493 KW - bayesian KW - commentary KW - hierarchical KW - model-structure KW - overview KW - smoothing N1 - A useful explanatory paper which clearly explains what Bayesian smoothing is supposed to do, and under what circumstances it may be used. ``Observable data can be conceptualized as `structure plus noise' with the role of analysis being to reveal the structure by stripping away the noise.'' (p.493) ``When data are not plentiful for all analytical units, noise can be reduced by “borrowing strength” or pooling information across analytical units. This requires a modeling framework that connects separate analytical units.'' (p.493) AU - Graham, P PY - 2008/05// UR - http://dx.doi.org/10.1097/EDE.0b013e31816b7859 ER -