Theriseofprobabilitytheorychangedthat.StatisticalinferencecompelsusinsteadtorelyonFortunaasaservantofMinerva,tousechanceanduncertaintytodiscoverreliableknowledge.Allflavorsofstatisticalinferencehavethismotivation.ButBayesiandataanalysisembracesitmostfully,byusingthelanguageofchancetodescribetheplausibilityofdifferentpossibilities.Therearemanywaystousetheterm“Bayesian.”Butmainlyitdenotesaparticularinterpretationofprobability.Inmodestterms,Bayesianinferenceisnomorethancountingthenumbersofwaysthingscanhappen,accordingtoourassumptions.Thingsthatcanhappenmorewaysaremoreplausible.Andsinceprobabilitytheoryisjustacalculusforcounting,thismeansthatwecanuseprobabilitytheoryasageneralwaytorepresentplausibility,whetherinreferencetocountableeventsintheworldorrathertheoreticalconstructslikeparameters.Onceyouacceptthisgambit,therestfollowslogically.Oncewehavedefinedourassumptions,Bayesianinferenceforcesapurelylogicalwayofprocessingthatinformationtoproduceinference.
2021/9/11 13:33:32
11.78MB
贝叶斯
统计
1