Version 0.79.0

abacusai/__init__.py

@@ -4,4 +4,4 @@
 from .streaming_client import StreamingClient
 
 
-__version__ = "0.78.5"
+__version__ = "0.79.0"

abacusai/api_class/dataset.py

@@ -18,3 +18,4 @@ class DocumentProcessingConfig(ApiClass):
     use_doctr: bool = False
     remove_watermarks: bool = None
     use_full_ocr: bool = None
+    layout_analysis: bool = False

abacusai/api_class/enums.py

@@ -35,6 +35,7 @@ class ProblemType(ApiEnum):
     DOCUMENT_VISUALIZATION = 'nlp_document_visualization'
     PERSONALIZATION = 'personalization'
     PREDICTIVE_MODELING = 'regression'
+    FINETUNED_LLM = 'finetuned_llm'
     FORECASTING = 'forecasting'
     CUSTOM_TRAINED_MODEL = 'plug_and_play'
     CUSTOM_ALGORITHM = 'trainable_plug_and_play'
@@ -387,6 +388,8 @@ class MonitorAlertType(ApiEnum):
     FEATURE_DRIFT = 'FeatureDrift'
     DATA_INTEGRITY_VIOLATIONS = 'DataIntegrityViolations'
     BIAS_VIOLATIONS = 'BiasViolations'
+    HISTORY_LENGTH_DRIFT = 'HistoryLengthDrift'
+    TARGET_DRIFT = 'TargetDrift'
 
 
 class FeatureDriftType(ApiEnum):
@@ -413,3 +416,9 @@ class BiasType(ApiEnum):
 
 class AlertActionType(ApiEnum):
     EMAIL = 'Email'
+
+
+class PythonFunctionType(ApiEnum):
+    FEATURE_GROUP = 'FEATURE_GROUP'
+    PLOTLY_FIG = 'PLOTLY_FIG'
+    STEP_FUNCTION = 'STEP_FUNCTION'

abacusai/api_class/model.py

@@ -48,14 +48,20 @@ class PersonalizationTrainingConfig(TrainingConfig):
         disable_transformer (bool): Disable training the transformer algorithm.
         disable_gpu (boo): Disable training on GPU.
         filter_history (bool): Do not recommend items the user has already interacted with.
+        action_types_exclusion_days (Dict[str, float]): Mapping from action type to number of days for which we exclude previously interacted items from prediction
+        session_dedupe_mins (float): Minimum number of minutes between two sessions for a user.
         max_history_length (int): Maximum length of user-item history to include user in training examples.
         compute_rerank_metrics (bool): Compute metrics based on rerank results.
         add_time_features (bool): Include interaction time as a feature.
         disable_timestamp_scalar_features (bool): Exclude timestamp scalar features.
         compute_session_metrics (bool): Evaluate models based on how well they are able to predict the next session of interactions.
         max_user_history_len_percentile (int): Filter out users with history length above this percentile.
         downsample_item_popularity_percentile (float): Downsample items more popular than this percentile.
+<<<<<<< HEAD
         allow_duplicate_action_types (List[str]): event types which will not be deduplicated.
+        use_user_id_feature (bool): Use user id as a feature in CTR models.
+=======
+>>>>>>> e5f12c159c (add config options for session dedupe and recommendation exclusion)
     """
     # top-level params
     objective: enums.PersonalizationObjective = dataclasses.field(default=None)
@@ -67,7 +73,6 @@ class PersonalizationTrainingConfig(TrainingConfig):
     target_action_types: List[str] = dataclasses.field(default=None)
     target_action_weights: Dict[str, float] = dataclasses.field(default=None)
     session_event_types: List[str] = dataclasses.field(default=None)
-    allow_duplicate_action_types: List[str] = dataclasses.field(default=None)
 
     # data split
     test_split: int = dataclasses.field(default=None)
@@ -92,6 +97,7 @@ class PersonalizationTrainingConfig(TrainingConfig):
 
     # prediction
     filter_history: bool = dataclasses.field(default=None)
+    action_types_exclusion_days: Dict[str, float] = dataclasses.field(default=None)
 
     # data distribution
     max_history_length: int = dataclasses.field(default=None)
@@ -101,6 +107,8 @@ class PersonalizationTrainingConfig(TrainingConfig):
     compute_session_metrics: bool = dataclasses.field(default=None)
     query_column: str = dataclasses.field(default=None)
     item_query_column: str = dataclasses.field(default=None)
+    use_user_id_feature: bool = dataclasses.field(default=None)
+    session_dedupe_mins: float = dataclasses.field(default=None)
 
     # outliers
     max_user_history_len_percentile: int = dataclasses.field(default=None)
@@ -156,7 +164,6 @@ class RegressionTrainingConfig(TrainingConfig):
         data_split_feature_group_table_name (str): Specify the table name of the feature group to export training data with the fold column.
         custom_loss_functions (List[str]): Registered custom losses available for selection.
         custom_metrics (List[str]): Registered custom metrics available for selection.
-
     """
     objective: enums.RegressionObjective = dataclasses.field(default=None)
     sort_objective: enums.RegressionObjective = dataclasses.field(default=None)

abacusai/api_class/monitor_alert.py

@@ -35,15 +35,47 @@ class FeatureDriftConditionConfig(AlertConditionConfig):
         feature_drift_type (FeatureDriftType): Feature drift type to apply the threshold on to determine whether a column has drifted significantly enough to be a violation.
         threshold (float): Threshold for when to consider a column to be in violation. The alert will only fire when the drift value is strictly greater than the threshold.
         minimum_violations (int): Number of columns that must exceed the specified threshold to trigger an alert.
+        feature_names (List[str]): List of feature names to monitor for this alert.
     """
     feature_drift_type: enums.FeatureDriftType = dataclasses.field(default=None)
     threshold: float = dataclasses.field(default=None)
     minimum_violations: int = dataclasses.field(default=None)
+    feature_names: List[str] = dataclasses.field(default=None)
 
     def __post_init__(self):
         self.alert_type = enums.MonitorAlertType.FEATURE_DRIFT
 
 
+@dataclasses.dataclass
+class TargetDriftConditionConfig(AlertConditionConfig):
+    """
+    Target Drift Condition Config for Monitor Alerts
+    Args:
+        feature_drift_type (FeatureDriftType): Target drift type to apply the threshold on to determine whether a column has drifted significantly enough to be a violation.
+        threshold (float): Threshold for when to consider the target column to be in violation. The alert will only fire when the drift value is strictly greater than the threshold.
+    """
+    feature_drift_type: enums.FeatureDriftType = dataclasses.field(default=None)
+    threshold: float = dataclasses.field(default=None)
+
+    def __post_init__(self):
+        self.alert_type = enums.MonitorAlertType.TARGET_DRIFT
+
+
+@dataclasses.dataclass
+class HistoryLengthDriftConditionConfig(AlertConditionConfig):
+    """
+    History Length Drift Condition Config for Monitor Alerts
+    Args:
+        feature_drift_type (FeatureDriftType): History length drift type to apply the threshold on to determine whether the history length has drifted significantly enough to be a violation.
+        threshold (float): Threshold for when to consider the history length  to be in violation. The alert will only fire when the drift value is strictly greater than the threshold.
+    """
+    feature_drift_type: enums.FeatureDriftType = dataclasses.field(default=None)
+    threshold: float = dataclasses.field(default=None)
+
+    def __post_init__(self):
+        self.alert_type = enums.MonitorAlertType.HISTORY_LENGTH_DRIFT
+
+
 @dataclasses.dataclass
 class DataIntegrityViolationConditionConfig(AlertConditionConfig):
     """
@@ -86,6 +118,8 @@ class _AlertConditionConfigFactory(_ApiClassFactory):
         enums.MonitorAlertType.FEATURE_DRIFT: FeatureDriftConditionConfig,
         enums.MonitorAlertType.DATA_INTEGRITY_VIOLATIONS: DataIntegrityViolationConditionConfig,
         enums.MonitorAlertType.BIAS_VIOLATIONS: BiasViolationConditionConfig,
+        enums.MonitorAlertType.TARGET_DRIFT: TargetDriftConditionConfig,
+        enums.MonitorAlertType.HISTORY_LENGTH_DRIFT: HistoryLengthDriftConditionConfig,
     }
 
 

abacusai/app_user_group.py

@@ -11,19 +11,21 @@ class AppUserGroup(AbstractApiClass):
             name (str): The name of the user group.
             userGroupId (str): The unique identifier of the user group.
             externalApplicationIds (list[str]): The ids of the external applications the group has access to.
+            invitedUserEmails (list[str]): The emails of the users invited to the user group who have not yet accepted the invite.
             users (User): The users in the user group.
     """
 
-    def __init__(self, client, name=None, userGroupId=None, externalApplicationIds=None, users={}):
+    def __init__(self, client, name=None, userGroupId=None, externalApplicationIds=None, invitedUserEmails=None, users={}):
         super().__init__(client, None)
         self.name = name
         self.user_group_id = userGroupId
         self.external_application_ids = externalApplicationIds
+        self.invited_user_emails = invitedUserEmails
         self.users = client._build_class(User, users)
 
     def __repr__(self):
-        repr_dict = {f'name': repr(self.name), f'user_group_id': repr(
-            self.user_group_id), f'external_application_ids': repr(self.external_application_ids), f'users': repr(self.users)}
+        repr_dict = {f'name': repr(self.name), f'user_group_id': repr(self.user_group_id), f'external_application_ids': repr(
+            self.external_application_ids), f'invited_user_emails': repr(self.invited_user_emails), f'users': repr(self.users)}
         class_name = "AppUserGroup"
         repr_str = ',\n  '.join([f'{key}={value}' for key, value in repr_dict.items(
         ) if getattr(self, key, None) is not None])
@@ -36,6 +38,6 @@ def to_dict(self):
         Returns:
             dict: The dict value representation of the class parameters
         """
-        resp = {'name': self.name, 'user_group_id': self.user_group_id,
-                'external_application_ids': self.external_application_ids, 'users': self._get_attribute_as_dict(self.users)}
+        resp = {'name': self.name, 'user_group_id': self.user_group_id, 'external_application_ids': self.external_application_ids,
+                'invited_user_emails': self.invited_user_emails, 'users': self._get_attribute_as_dict(self.users)}
         return {key: value for key, value in resp.items() if value is not None}