rockfish.labs.recommender
Recommender
A recommendation engine.
Functions
from_dataset(dataset: LocalDataset, col_resembles_timestamps: str = None, dependent_fields: list[str] = None, session_key: str = None, keep_session_keys: bool = False, metadata: list[str] = None, other_categorical: list[str] = None, training_cost: TrainingCostValue = TrainingCostValue.LOW, model: ModelType = ModelType.UNSPECIFIED, run_pii_detection: bool = True)
classmethod
Initializes from a dataset.
Parameters:
| Name | Type | Description | Default |
|---|---|---|---|
dataset
|
LocalDataset
|
Dataset. |
required |
col_resembles_timestamps
|
str
|
Name of the column that resembles timestamps. |
None
|
dependent_fields
|
list[str]
|
Fields with complete association/dependency. |
None
|
session_key
|
str
|
Name of column used as the session key. |
None
|
keep_session_keys
|
bool
|
Whether to keep the session keys after generation. |
False
|
metadata
|
list[str]
|
Metadata fields for sessions. Must be categorical. |
None
|
other_categorical
|
list[str]
|
Categorical measurement fields. |
None
|
training_cost
|
TrainingCostValue
|
Training cost requirement. |
LOW
|
model
|
ModelType
|
Model/architecture to be used. |
UNSPECIFIED
|
run_pii_detection
|
bool
|
Whether to run PII detection or skip it. |
True
|
recommendations() -> Recommendation
Returns the recommendations.