This function requests the creation of a new model in the DataRobot modeling project defined by the project parameter. The function also allows the user to specify alternatives to the project default for featurelist, samplePct, and scoringType. This function returns an integer modelJobId value, which can be used by the GetModelFromJobId function to return the full model object.
RequestNewModel(
project,
blueprint,
featurelist = NULL,
samplePct = NULL,
trainingRowCount = NULL,
scoringType = NULL,
monotonicIncreasingFeaturelistId = NULL,
monotonicDecreasingFeaturelistId = NULL
)An integer value that can be used as the modelJobId parameter in subsequent calls to the GetModelFromJobId function.
character. Either (1) a character string giving the unique alphanumeric identifier for the project, or (2) a list containing the element projectId with this identifier.
list. A list with at least the following two elements: blueprintId and projectId. Note that the individual elements of the list returned by ListBlueprints are admissible values for this parameter.
list. A list that contains the element featurelistId that specifies the featurelist to be used in building the model; if not specified (i.e., for the default value NULL), the project default (Informative Features) is used.
numeric. The percentage of the training dataset to be used in building the new model; if not specified (i.e., for the default value NULL), the maxTrainPct value for the project is used. Value should be between 0 and 100.
integer. The number of rows to use to train the requested model.
character. String specifying the scoring type; default is validation set scoring, but cross-validation averaging is also possible.
character. Optional. The id of the featurelist
that defines the set of features with a monotonically increasing relationship to the
target. If NULL (default), the default for the project will be used (if any).
Note that currently there is no way to create a model without monotonic constraints
if there was a project-level default set. If desired, the featurelist itself can
also be passed as this parameter.
character. Optional. The id of the featurelist
that defines the set of features with a monotonically decreasing relationship to the
target. If NULL, the default for the project will be used (if any). If empty
(i.e., ""), no such constraints are enforced. Also, if desired, the featurelist
itself can be passed as this parameter.
Motivation for this function is the fact that some models - e.g., very complex machine learning models fit to large datasets - may take a long time to complete. Splitting the model creation request from model retrieval in these cases allows the user to perform other interactive R session tasks between the time the model creation/update request is made and the time the final model is available.
Either `sample_pct` or `training_row_count` can be used to specify the amount of data to use, but not both. If neither are specified, a default of the maximum amount of data that can safely be used to train any blueprint without going into the validation data will be selected. In smart-sampled projects, `samplePct` and `trainingRowCount` are assumed to be in terms of rows of the minority class.
Note : For datetime partitioned projects, use RequestNewDatetimeModel instead
if (FALSE) {
projectId <- "59a5af20c80891534e3c2bde"
blueprints <- ListBlueprints(projectId)
blueprint <- blueprints[[1]]
RequestNewModel(projectId, blueprint)
}
Run the code above in your browser using DataLab