Represents the complete definition of a model in the context of the GeneXus Cognitive API for custom models.
- Name: VarChar(64) -- Model's name
- Dataset (structure)
- Loader: ObjectName, GeneXus -- Fully qualified object name that generates training data samples.
- Option (Structure)
- Batch: Numeric(4.0) -- Describes how many elements will be processed time by time.
- Size: (structure) -- Proportion reservation in range [0,1].
- Training: Numeric(5.3)
- Testing: Numeric(5.3)
- Seed: Numeric(4.0) -- Useful when you want to repeat an experiment (same random splitting).
- Input (structure)
- Features (collection) -- Defines every data input by name and type (i.e. every column names/types).
- Output (structure)
- Dataset.Loader field must be set with the fully-qualified name of your dataset 'generator' object (Procedure or Data Provider).
- Features' field count in the Input field must be equal to every Data.Input Features' field count of your data-set.
- Features' field in Definition.Input and Data.Input must be correlative in order to describe the name and type of every feature of your data item (i.e. the k-th element of Data.Input.Feature should have the Type of the k-th element of Definition.Input.Feature).
- If Data.Input.Features field has more items than Definition.Input.Features field, remain items in Data.Input.Features will be ignored.
- If Data.Input.Features field has fewer items than Definition.Input.Features, means you have undefined Data.Input.Features and will produce an error.
- When Name field is omitted in Data.Input.Features' item, the column name will be the Type field name, enumerated in order of appearance (e.g. "MEDIA_1"). In order to data summary being human readable, it is recommended to do not omit the Name field.
This data type is available as of GeneXus 16 upgrade 6.