Package: heimdall 1.0.717
Eduardo Ogasawara
heimdall:Drift Adaptable Models
By analyzing streaming datasets, it is possible to observe significant changes in the data distribution or models' accuracy during their prediction (concept drift). The goal of 'heimdall' is to measure when concept drift occurs. The package makes available several state-of-the-art methods. It also tackles how to adapt models in a nonstationary context. Some concept drifts methods are described in Tavares (2022) <doi:10.1007/s12530-021-09415-z>.
Authors:
heimdall_1.0.717.tar.gz
heimdall_1.0.717.tar.gz(r-4.5-noble)heimdall_1.0.717.tar.gz(r-4.4-noble)
heimdall_1.0.717.tgz(r-4.4-emscripten)heimdall_1.0.717.tgz(r-4.3-emscripten)
heimdall.pdf |heimdall.html✨
heimdall/json (API)
| # Installheimdall in R: |
| install.packages('heimdall',repos = c('https://cran.r-universe.dev', 'https://cloud.r-project.org')) |
Bug tracker:https://github.com/cefet-rj-dal/heimdall/issues
- st_drift_examples - Synthetic time series for concept drift detection
Last updated 5 days agofrom:3cd2967a47
Exports:dfr_adwindfr_cusumdfr_ddmdfr_ecdddfr_eddmdfr_hddmdfr_inactivedfr_kldistdfr_kswindfr_mcdddfr_page_hinkleydfr_passivedist_baseddriftererror_basedmetricmt_fscoremt_precisionmt_recallmulti_criteriareset_statestealthyupdate_state
Dependencies:bitopscaretcaToolsclasscliclockclustercodetoolscolorspacecpp11curldaltoolboxdata.tabledbscandiagramdigestdplyre1071ellipsiselmNNRcppfansifarverFNNforeachforecastfracdifffuturefuture.applygenericsggplot2globalsgluegowergplotsgtablegtoolshardhathereipredisobanditeratorsjsonliteKernelKnnKernSmoothlabelinglatticelavalifecyclelistenvlmtestlubridatemagrittrMASSMatrixmgcvMLmetricsModelMetricsmunsellnlmennetnumDerivparallellypillarpkgconfigplyrpngpROCprodlimprogressrproxypurrrquadprogquantmodR6randomForestrappdirsRColorBrewerRcppRcppArmadilloRcppTOMLrecipesreshapereshape2reticulaterlangROCRrpartrprojrootscalesshapeSQUAREMstringistringrsurvivaltibbletidyrtidyselecttimechangetimeDatetreetseriesTTRtzdburcautf8vctrsviridisLitewithrxtszoo
Readme and manuals
Help Manual
| Help page | Topics |
|---|---|
| ADWIN method | dfr_adwin |
| Cumulative Sum for Concept Drift Detection (CUMSUM) method | dfr_cusum |
| Adapted Drift Detection Method (DDM) method | dfr_ddm |
| Adapted EWMA for Concept Drift Detection (ECDD) method | dfr_ecdd |
| Adapted Early Drift Detection Method (EDDM) method | dfr_eddm |
| Adapted Hoeffding Drift Detection Method (HDDM) method | dfr_hddm |
| Inactive dummy detector | dfr_inactive |
| KL Distance method | dfr_kldist |
| KSWIN method | dfr_kswin |
| Mean Comparison Distance method | dfr_mcdd |
| Adapted Page Hinkley method | dfr_page_hinkley |
| Passive dummy detector | dfr_passive |
| Distribution Based Drifter sub-class | dist_based |
| Drifter | drifter |
| Error Based Drifter sub-class | error_based |
| Process Batch | fit.drifter |
| Metric | metric |
| FScore Calculator | mt_fscore |
| Precision Calculator | mt_precision |
| Recall Calculator | mt_recall |
| Multi Criteria Drifter sub-class | multi_criteria |
| Reset State | reset_state |
| Synthetic time series for concept drift detection | st_drift_examples |
| Stealthy | stealthy |
| Update State | update_state |
