C# 클래스 QACExperimenter.Approaches.noSGDLRMNTB

Linear regression over multiple non-overlapping non temporal buckets. The linear regression model is trained using stochastic gradient descent (SGD) during the run.
상속: BaseApproach
파일 보기 프로젝트 열기: stewhir/recent-robust-qac

공개 메소드들

메소드 설명
CreateAutoCompletionList ( IEnumerable allPrefixEntries, int defaultMinCutOff = 1 ) : AutoCompletionList
noSGDLRMNTB ( int multipleNs, int multipleMaxSingleQueryN, int trainAfterNQueriesForPrefix, string queryLogFile, int autoCompleteAfterNChars, StandardEvalOutput evalOutput, PrefixProfile queryPrefixProfile, int baseNtbSize = 200 ) : System

MultipleNs must be specified in ascending order for the NTBs

보호된 메소드들

메소드 설명
AutoCompleteQuery ( System.DateTime queryTime, string partialQuery, string fullQuery ) : AutoCompletionList

Autocomplete a query after n characters

비공개 메소드들

메소드 설명
BuildFeaturePackage ( BaseIndexEntry prefixEntry, string partialQuery, bool includeQueryCountsFeature = false ) : FeaturePackage

Build the feature package including NTB likelihoods for the specified query. Doesn't include the target likelihood.

BuildTrainingPackage ( string partialQuery, bool isFirstTrainingPackageForPrefix ) : TrainingPackage

Build the training package for a query prefix

MakePrediction ( BaseIndexEntry indexEntry, string partialQuery ) : double

Make prediction using linear regression

UseTrainingPackage ( TrainingPackage trainingPackage ) : void

Apply the training package to the model

메소드 상세

AutoCompleteQuery() 보호된 메소드

Autocomplete a query after n characters
protected AutoCompleteQuery ( System.DateTime queryTime, string partialQuery, string fullQuery ) : AutoCompletionList
queryTime System.DateTime
partialQuery string
fullQuery string
리턴 AutoCompletionList

CreateAutoCompletionList() 공개 메소드

public CreateAutoCompletionList ( IEnumerable allPrefixEntries, int defaultMinCutOff = 1 ) : AutoCompletionList
allPrefixEntries IEnumerable
defaultMinCutOff int
리턴 AutoCompletionList

noSGDLRMNTB() 공개 메소드

MultipleNs must be specified in ascending order for the NTBs
public noSGDLRMNTB ( int multipleNs, int multipleMaxSingleQueryN, int trainAfterNQueriesForPrefix, string queryLogFile, int autoCompleteAfterNChars, StandardEvalOutput evalOutput, PrefixProfile queryPrefixProfile, int baseNtbSize = 200 ) : System
multipleNs int
multipleMaxSingleQueryN int
trainAfterNQueriesForPrefix int How many queries to observe between training the ML model (i.e., predict queries in this window of N - OR: 'the prediction horizon'). Note the model won't start training until after (largest NTB size) + (trainAfter parameter) has been reached
queryLogFile string
autoCompleteAfterNChars int
evalOutput QACExperimenter.Evaluation.StandardEvalOutput
queryPrefixProfile PrefixProfile
baseNtbSize int
리턴 System