Package: heimdall 1.4.0

Eduardo Ogasawara

heimdall: Drift Adaptable Models

In streaming data analysis, it is crucial to detect significant shifts in the data distribution or the accuracy of predictive models over time, a phenomenon known as concept drift. The package aims to identify when concept drift occurs and provide methodologies for adapting models in non-stationary environments. It offers a range of state-of-the-art techniques for detecting concept drift and maintaining model performance. Additionally, the package provides tools for adapting models in response to these changes, ensuring continuous and accurate predictions in dynamic contexts. Methods for concept drift detection are described in Tavares (2022) <doi:10.1007/s12530-021-09415-z>.

Authors:Lucas Tavares [aut], Leonardo Carvalho [aut], Rodrigo Machado [aut], Diego Carvalho [ctb], Esther Pacitti [ctb], Fabio Porto [ctb], Eduardo Ogasawara [aut, ths, cre], CEFET/RJ [cph]

heimdall_1.4.0.tar.gz
heimdall_1.4.0.zip(r-4.7-any)heimdall_1.4.0.zip(r-4.6-any)heimdall_1.4.0.zip(r-4.5-any)
heimdall_1.4.0.tgz(r-4.6-any)heimdall_1.4.0.tgz(r-4.5-any)
heimdall_1.4.0.tar.gz(r-4.7-any)heimdall_1.4.0.tar.gz(r-4.6-any)
heimdall_1.4.0.tgz(r-4.6-emscripten)
manual.pdf |manual.html✨
DESCRIPTION |NEWS
card.svg |card.png
heimdall/json (API)

# Install 'heimdall' in R:
install.packages('heimdall', repos = c('https://cefet-rj-dal.r-universe.dev', 'https://cloud.r-project.org'))

Bug tracker:https://github.com/cefet-rj-dal/heimdall/issues

Datasets:

On CRAN:

Conda:

5.18 score 2 stars 579 downloads 79 exports 82 dependencies

Last updated from:d8fe9b91ff. Checks:9 OK. Indexed: yes.

TargetResultTimeFilesSyslog
linux-develOK345
source / vignettesOK246
linux-releaseOK352
macos-releaseOK305
macos-oldrelOK243
windows-develOK318
windows-releaseOK318
windows-oldrelOK293
wasm-releaseOK148

Exports:cosine_scorecpd_calibratecpd_ensemblecpd_ensemble_fitcpd_scoresdde_curvedde_feature_driftdde_scoredfr_adwindfr_aedddfr_cosinedfr_cusumdfr_ddedfr_ddmdfr_ecdddfr_eddmdfr_eikmeansdfr_ensembledfr_hddmdfr_inactivedfr_indiddfr_kldistdfr_kswindfr_lbdddfr_llr2dfr_mcddfr_mcdddfr_mmqtdfr_multi_criteriadfr_onlinebcpdfr_padddfr_page_hinkleydfr_passivedfr_protondfr_psidfr_qtsdfr_studddfr_wwaggrdist_baseddrifterdsg_agrawaldsg_driftdsg_mixeddsg_mmd_testdsg_seadsg_sinedsg_sourcedsg_source_datadsg_source_regionsdsg_streamdsg_takedummyeikmeanseikmeans_testensemble_voteerror_basedindidindid_predictmetricmmqtmmqt_testmt_accuracymt_fscoremt_precisionmt_recallmt_rocaucmv_dist_basednrm_basenrm_memoryonline_bcppsireset_statespbnspbn_loglikstealthyts2vectscp2update_statewwaggr

Dependencies:arulesarulesSequencescaretclasscliclockclustercodetoolscpp11daltoolboxdata.tabledbscandiagramdplyre1071farverFNNforeachfuturefuture.applygenericsggplot2globalsgluegowergtablehardhatipredisobanditeratorsKernSmoothlabelinglatticelavalifecyclelistenvlubridatemagrittrMASSMatrixmclustnlmennetnumDerivparallellypillarpkgconfigplyrpROCprodlimprogressrproxypurrrR6randomForestRColorBrewerRcpprecipesreshapereshape2rlangrpartS7scalesshapesparsevctrsSQUAREMstringistringrsurvivaltibbletidyrtidyselecttimechangetimeDatetreetzdbutf8vctrsviridisLitewithryardstick

Adapting models to concept drift
A stream with real concept drift | Training and testing batch by batch | Choosing the detector

Last update: 2026-10-16
Started: 2026-10-15

Configuring detectors
Reading the drifter output | Calibrating a threshold | Sliding and anchored windows | Speed on long streams

Last update: 2026-10-16
Started: 2026-10-16

Getting started with heimdall
Real and virtual concept drift | The streaming contract | Virtual drift: monitoring a feature | Real drift: monitoring the errors of a model | Processing a whole batch | Combining detectors | Baselines | Next steps

Last update: 2026-10-16
Started: 2026-10-15

Readme and manuals

Help Manual

Help pageTopics
Average cosine similarity to a baseline of embeddingscosine_score
Calibration of change point modelscpd_calibrate
Ensembles of change point modelscpd_ensemble cpd_ensemble_fit
Change point scores of a sequence modelcpd_scores
DDE probability curvedde_curve
Feature-wise drift tests between two windowsdde_feature_drift
DDE drift scoresdde_score
ADWIN methoddfr_adwin
Autoencoder-Based Drift Detection methoddfr_aedd
Cosine similarity drift detector for embeddingsdfr_cosine
Cumulative Sum for Concept Drift Detection (CUSUM) methoddfr_cusum
Domino drift effect (DDE) detectordfr_dde
Adapted Drift Detection Method (DDM) methoddfr_ddm
Adapted EWMA for Concept Drift Detection (ECDD) methoddfr_ecdd
Adapted Early Drift Detection Method (EDDM) methoddfr_eddm
EI-kMeans stream detectordfr_eikmeans
Ensemble drift detector (CEDD and FEDD)dfr_ensemble
Adapted Hoeffding Drift Detection Method (HDDM) methoddfr_hddm
Inactive dummy detectordfr_inactive
InDiD stream detectordfr_indid
KL Distance methoddfr_kldist
KSWIN methoddfr_kswin
Levene Based Drift Detection Method methoddfr_lbdd
LLR2 drift detectordfr_llr2
Maximum Concept Discrepancy drift detector (MCD-DD)dfr_mcd
Mean Comparison Distance methoddfr_mcdd
Multimodal QuantTree stream detectordfr_mmqt
Multi Criteria Drifter sub-classdfr_multi_criteria
onlineBcp stream detectordfr_onlinebcp
Parallel Activations Drift Detector (PADD)dfr_padd
Adapted Page Hinkley methoddfr_page_hinkley
Passive dummy detectordfr_passive
PROTON: prototype-based test-time online OOD detectiondfr_proton
PSI stream detectordfr_psi
Quadtree spatial mapping drift detector (QTS)dfr_qts
Student-Teacher Unsupervised Drift Detection (STUDD)dfr_studd
WWAggr ensemble drift detectordfr_wwaggr
Distribution Based Drifter sub-classdist_based
Drifterdrifter
Induce a drift in a benchmark sourcedsg_drift
Kernel two-sample test (MMD)dsg_mmd_test
Drift benchmark sourcedsg_source
Benchmark source from a real-world streamdsg_source_data
Benchmark source from a synthetic generatordsg_source_regions
Generate a benchmark stream with controlled driftdsg_stream
Synthetic benchmark sourcesdsg_agrawal dsg_mixed dsg_sea dsg_sine dsg_synthetic
Draw samples from a drift benchmark sourcedsg_take
Dummy Drifter sub-classdummy
Equal intensity k-means space partitioning (EI-kMeans)eikmeans
Test a sample with an EI-kMeans partitioningeikmeans_test
Ensemble vote over detection seriesensemble_vote
Error Based Drifter sub-classerror_based
Process Batchfit.drifter
InDiD change point detection modelindid
Probabilities of change of an InDiD modelindid_predict
Metricmetric
Multimodal QuantTree (MMQT)mmqt
Test batches with a Multimodal QuantTreemmqt_test
Accuracy Calculatormt_accuracy
FScore Calculatormt_fscore
Precision Calculatormt_precision
Recall Calculatormt_recall
ROC AUC Calculatormt_rocauc
Multivariate Distribution Based Drifter sub-classmv_dist_based
Normalizer base classnrm_base
Memory Normalizernrm_memory
Online Bayesian change point detection (onlineBcp)online_bcp
Population Stability Index (PSI)psi
Reset Statereset_state
Semiparametric Bayesian networkspbn
Log-likelihood of each node of a semiparametric Bayesian networkspbn_loglik
Synthetic time series for concept drift detectionst_drift_examples
Stealthystealthy
SN-TS2Vec change point modelts2vec
TS-CP2 change point modeltscp2
Update Stateupdate_state
WWAggr and naive aggregations of ensemble scoreswwaggr