ColumnInformation Class

Definition

Information about the columns in a dataset.

public sealed class ColumnInformation
type ColumnInformation = class
Public NotInheritable Class ColumnInformation
Inheritance
ColumnInformation

Remarks

Contains information about the purpose of each column in the dataset. For instance, it enumerates the dataset columns that AutoML should treat as categorical, the columns AutoML should ignore, which column is the label, etc.

ColumnInformation can be fed to the AutoML API when running an experiment. See for example.

Constructors

ColumnInformation()

Properties

CategoricalColumnNames

The dataset columns that are categorical.

ExampleWeightColumnName

The dataset column to use for example weight.

GroupIdColumnName

The dataset column to use as a group ID for computation in a Ranking Task. If a SamplingKeyColumnName is provided, then it should be the same as this column.

IgnoredColumnNames

The dataset columns that AutoML should ignore.

ImagePathColumnNames

The dataset columns that are image paths.

ItemIdColumnName

The dataset column to use as a item ID for computation.

LabelColumnName

The dataset column to use as the label.

NumericColumnNames

The dataset columns that are numeric.

SamplingKeyColumnName

The dataset column to use for grouping rows. If two examples share the same sampling key column name, they are guaranteed to appear in the same subset (train or test). This can be used to ensure no label leakage from the train to the test set. If null, no row grouping will be performed.

TextColumnNames

The dataset columns that are text.

UserIdColumnName

The dataset column to use as a user ID for computation.

Applies to