Skip to main content

Units

All registered Unit components in DashAI.

NameDescription
ApplyConverterUnitFit one converter on the scoped data and transform the same dataset.
ApplyCutPreprocessingUnitPublish the exogenous variables each cut of a backtest reads.
ApplyDatasetSchemaUnitRename the dataset's columns and cast them to their declared types.
ApplyDocumentPipelineUnitRun the configured converter pipeline and republish the document.
ApplySessionPreprocessingUnitPublish the rows as the model was trained to see them.
BacktestForecasterUnitForecast every origin of a backtest, searching first if there is a search.
BuildForecasterUnitInstantiate an unfitted forecaster, with the season its grid implies.
BuildGlobalExplainerUnitInstantiate a global explainer over an already trained model.
BuildLocalExplainerUnitInstantiate a local explainer over an already trained model.
BuildManualInputUnitBuild the dataset to predict on from values the user typed in.
BuildModelUnitInstantiate an untrained model bound to its data and metrics.
ComputeDatasetMetadataUnitFill in the metadata a dataset carries in its splits mapping.
EvaluateModelToArtifactUnitCompute a trained model's metrics and publish them as data.
EvaluateModelUnitCompute the final metrics of a trained model, once per split.
FitConverterUnitFit a converter on the dataset in the context and publish it fitted.
FitFinalForecasterUnitFit the model a run keeps, with the best parameters, on the whole series.
FitModelOverFoldsUnitFit a model once per fold, then once more on everything the folds used.
FitModelOverNestedFoldsUnitScore every outer fold with a search that never saw it.
FitModelUnitTrain a model, running a hyperparameter search when there is one to run.
FitWithoutTargetUnitFit a model on the features of every row, with no target column.
ForecastFutureUnitForecast the steps of the future rows, reading their exogenous values.
ForecastHorizonUnitForecast the steps after a trained series, and name their dates.
GenerateGlobalExplanationUnitExplain the model over the whole dataset and pickle the result.
GenerateLocalExplanationUnitExplain a selection of instances and pickle the result.
InferDatasetTypesUnitPublish a type declaration for every column of the dataset.
LoadDatafileDatasetUnitParse a dataset out of a download that already completed.
LoadDatasetUnitLoad a dataset from disk into the execution context.
LoadRunModelUnitRestore a run's trained model for an explanation.
LoadTrainedModelUnitRebuild a trained model from the run that produced it.
LoadTrainingDatasetUnitLoad the dataset a model was trained on, under a key of its own.
LoadUploadedDatasetUnitParse a file or URL into a dataset with the chosen reader.
ParseDocumentUnitParse source_path into a document with the chosen loader.
PredictUnitPredict with a trained model and decode the result into labels.
PrepareAndFoldUnitValidate a dataset against a task and carve it into cross-validation folds.
PrepareAndSplitUnitValidate a dataset against a task and split it once into three partitions.
PrepareExplanationDataUnitRebuild the exact train/test/val split the run was trained on.
PrepareFutureUnitCheck the rows a user brings for the steps after the series.
PrepareSeriesUnitTurn a dataset into a series on its grid, and cut it into origins.
PrepareWithoutTargetUnitPrepare a dataset for a task that declares no target column.
RunExplorationUnitInstantiate an explorer component and run it over the dataset.
SaveDatasetToPathUnitWrite the dataset to a destination given in the configuration.
SaveDatasetUnitWrite the dataset back to the path it was loaded from.
SaveDocumentToPathUnitWrite the document to a destination given in the configuration.
SaveExplorationUnitWrite an exploration result to disk and publish where it landed.
SaveForecastsPlotUnitDraw the forecasts of every origin and write the figures beside the model.
SaveForecastsUnitWrite the forecasts of every origin to a JSON file beside the model.
SaveModelUnitWrite a trained model under the runs directory, in its own subdirectory.
SavePredictionUnitWrite the predicted column next to the data it was predicted from.
ScoreClustersUnitScore a fitted clustering over every row of the features it was fitted on.
ScoreForecastsUnitScore each origin's forecasts into the rows a run's metrics are.
StoreDatafileDocumentSourceUnitCopy a file out of a completed hub download and hash it.
StoreUploadedDocumentSourceUnitCopy an uploaded file into a directory of its own and hash it.
TransformDatasetUnitApply an already fitted converter to the dataset in the context.