Please wait
{{ preTrainStatus.nbModelsPreGS | pluralize : getModelStr() : getModelStr(true) }} will be queued, training trained on {{ preTrainStatus.nbInputFeatures }} {{ 'feature' | plurify: preTrainStatus.nbInputFeatures }} with {{ preTrainStatus.nbInputFeatures }} {{ ' external feature' | plurify: preTrainStatus.nbInputFeatures }} using {{ mlTaskDesign.splitParams.nFolds || 'k' }}-fold cross-validation using custom cross-validation for hyper parameters optimization
An estimated total of {{preTrainStatus.estimatedTotalFits}} {{ 'estimator' | plurify: preTrainStatus.estimatedTotalFits }} will be evaluated. The hyperparameters search for each eligible model {{preTrainStatus.partitionedModelEnabled ? "partition": ""}} will run for at most {{preTrainStatus.timeout | pluralize: 'minute' : 'minutes'}} and up to {{ preTrainStatus.estimatedTotalFits }} {{ 'estimator' | plurify: preTrainStatus.estimatedTotalFits }} will be evaluated. The hyperparameters search for each eligible model {{preTrainStatus.partitionedModelEnabled ? "partition": ""}} will run for {{preTrainStatus.timeout | pluralize: 'minute' : 'minutes'}}.
At most {{preTrainStatus.maxConcurrentK8sContainers}} kubernetes containers will run concurrently. GPU will be used for {{usedGpuCapabilitiesWithGpuOn.length}} of {{usedGpuCapabilities.length}} possible steps.

Session Name & Description

Train & test

New {{useExtracts() ? 'extracts' : 'train and test sets'}} will be computed according to your settings
Re-using {{useExtracts() ? 'extracts' : 'sets'}} computed on {{splitStatus.generationDate|date:'yyyy/MM/dd-HH:mm'}}.

Train set

{{splitStatus.trainRows}} rows

Test set

{{splitStatus.testRows}} rows

Extract

{{splitStatus.fullRows}} rows
Note that the inputs may have been modified since.
Train ({{splitStatus.trainRows}} rows) and test ({{splitStatus.testRows}} rows) sets Extract ({{splitStatus.fullRows}} rows) set computed on {{splitStatus.generationDate|date:'yyyy/MM/dd-HH:mm'}} will be used.
Dataset {{splitStatus.modifiedDataset}} was updated since (on {{splitStatus.datasetModificationDate|date:'yyyy/MM/dd-HH:mm'}})
With this ML backend, train & test are always recomputed

Messages

The following algorithms require at least one generated feature:
  • - {{alg}}
They will fail during training.
You can configure generated features under the 'Feature generation' tab.
The following algorithms are not compatible with the sample weights option:
  • - {{alg}}
The variable {{mlTaskDesign.weight.sampleWeightVariable}} will be ignored while training these algorithms, but hyper-parameters optimization and model evaluation will still take the sample weight into account.
The following algorithms are not compatible with external features:
  • - {{ alg.name }}
The external features will be ignored while training these algorithms.
The following statistical algorithms might be slow to train on multiple time series as they train one model per time series:
  • - {{ alg.name }}
If you have many time series, it is recommended to use Deep Learning algorithms instead, to train a single model for all time series.
GPU has been enabled, but each available step has been disabled. Re-enable the steps in the runtime environment settings to allow execution on GPU.
  • {{message.title}}

    {{message.details}}
  • MLLib is deprecated

    MLLib will soon be removed.