{"id":2530,"date":"2018-05-07T07:18:27","date_gmt":"2018-05-07T01:18:27","guid":{"rendered":"https:\/\/www.isrt.ac.bd\/?post_type=tribe_events&p=2530"},"modified":"2018-05-07T07:18:47","modified_gmt":"2018-05-07T01:18:47","slug":"seminar-on-tuesday-april-8-at-2-pm","status":"publish","type":"tribe_events","link":"https:\/\/isrt.ac.bd\/event\/seminar-on-tuesday-april-8-at-2-pm\/","title":{"rendered":"Seminar on Tuesday, April 8 at 2 pm"},"content":{"rendered":"
Title<\/span>:\u00a0Approximation of the Formal Bayesian Model Comparison using the Extended Conditional Predictive Ordinatecriterion
\n
<\/span><\/div>\n
Speaker<\/span>: Md. Rashedul Hoque, Assistant Professor, ISRT<\/span><\/span><\/div>\n
\u00a0<\/span><\/div>\n
Venue and Time<\/span>: ISRT seminar room, 2 pm<\/span><\/span><\/div>\n
<\/div>\n
Abstract:\u00a0<\/span>The optimal method for Bayesian model comparison is the formal Bayes factor (BF), according\u00a0to decision theory. The formal BF is computationally troublesome for more complex models.\u00a0If predictive distributions under the competing models do not have a closed form, a cross-validation\u00a0idea, called the conditional predictive ordinate (CPO) criterion can be used. In\u00a0the cross-validation sense,\u00a0<\/span>this is a \u2018leave-out one\u2019 approach. CPO can be calculated directly\u00a0from the Monte Carlo (MC) outputs, and the resulting Bayesian model comparison is called\u00a0the pseudo-Bayes factor (PBF). We can get closer to the formal Bayesian model comparison\u00a0by increasing the \u2018leave-out size\u2019, and at \u2018leave-out all\u2019 we recover the formal BF. But, the\u00a0MC error increases with increasing \u2018leave-out size\u2019. In this study, we examine this for linear\u00a0and logistic regression models.<\/div>\n
Our study reveals that the Bayesian model comparison can favor a different model for PBFcompared to BF when comparing two close linear models. So, larger \u2018leave-out sizes\u2019 are\u00a0preferred which provide result close to the optimal BF. On the other hand, MC samples\u00a0based formal Bayesian model comparisons are computed with more MC error for increasing\u00a0\u2018leave-out sizes\u2019; this is observed by comparing with the available closed-form results. Still,\u00a0considering a reasonable error, we can use \u2018leave-out size\u2019 more than one instead of fixing\u00a0it at one. These findings can be extended to logistic models where closed form solution is
\nunavailable.<\/span><\/div>\n","protected":false},"excerpt":{"rendered":"

Title:\u00a0Approximation of the Formal Bayesian Model Comparison using the Extended Conditional Predictive Ordinatecriterion Speaker: Md. Rashedul Hoque, Assistant Professor, ISRT \u00a0 Venue and Time: ISRT seminar room, 2 pm Abstract:\u00a0The … [ Read More ]<\/a><\/p>\n","protected":false},"author":5,"featured_media":0,"template":"","meta":{"_acf_changed":false,"nf_dc_page":"","_monsterinsights_skip_tracking":false,"_monsterinsights_sitenote_active":false,"_monsterinsights_sitenote_note":"","_monsterinsights_sitenote_category":0,"_tribe_events_status":"","_tribe_events_status_reason":"","footnotes":""},"tags":[],"tribe_events_cat":[15],"class_list":["post-2530","tribe_events","type-tribe_events","status-publish","hentry","tribe_events_cat-seminar","cat_seminar"],"acf":[],"_links":{"self":[{"href":"https:\/\/isrt.ac.bd\/wp-json\/wp\/v2\/tribe_events\/2530","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/isrt.ac.bd\/wp-json\/wp\/v2\/tribe_events"}],"about":[{"href":"https:\/\/isrt.ac.bd\/wp-json\/wp\/v2\/types\/tribe_events"}],"author":[{"embeddable":true,"href":"https:\/\/isrt.ac.bd\/wp-json\/wp\/v2\/users\/5"}],"version-history":[{"count":1,"href":"https:\/\/isrt.ac.bd\/wp-json\/wp\/v2\/tribe_events\/2530\/revisions"}],"predecessor-version":[{"id":2531,"href":"https:\/\/isrt.ac.bd\/wp-json\/wp\/v2\/tribe_events\/2530\/revisions\/2531"}],"wp:attachment":[{"href":"https:\/\/isrt.ac.bd\/wp-json\/wp\/v2\/media?parent=2530"}],"wp:term":[{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/isrt.ac.bd\/wp-json\/wp\/v2\/tags?post=2530"},{"taxonomy":"tribe_events_cat","embeddable":true,"href":"https:\/\/isrt.ac.bd\/wp-json\/wp\/v2\/tribe_events_cat?post=2530"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}