nannyml.io.db.entities module

Contains the definitions of the database entities that map directly to the underlying table definitions.

Every Result class has a matching Entity class, which implies that each calculator or estimator will export its results into a specific table.

class nannyml.io.db.entities.CBPEPerformanceMetric(*, id: Optional[int] = None, model_id: Optional[int] = None, run_id: int = None, start_timestamp: datetime, end_timestamp: datetime, timestamp: datetime, metric_name: str, value: float, alert: bool, upper_threshold: Optional[float] = None, lower_threshold: Optional[float] = None)[source]

Bases: Metric

Represents results of the CBPE estimator.

Stored in the cbpe_performance_metrics table.

Create a new model by parsing and validating input data from keyword arguments.

Raises ValidationError if the input data cannot be parsed to form a valid model.

alert: bool: Indicates if the method raised an alert for this Chunk

end_timestamp: datetime: The end datetime of the Chunk

id: Optional[int]: The technical identifier for this database row

lower_threshold: Optional[float]: The lower alerting threshold value

metric_name: str: The name of the method being calculated, e.g. jensen_shannon or chi2

model_id: Optional[int]: Foreign key pointing to a record in the model table

run_id: int: Foreign key pointing to a record in the run table

start_timestamp: datetime: The start datetime of the Chunk

timestamp: datetime: The ‘’center’’ timestamp of the Chunk, i.e. the mean of the start and end timestamps

upper_threshold: Optional[float]: The upper alerting threshold value

value: float: The value returned by the method

class nannyml.io.db.entities.DLEPerformanceMetric(*, id: Optional[int] = None, model_id: Optional[int] = None, run_id: int = None, start_timestamp: datetime, end_timestamp: datetime, timestamp: datetime, metric_name: str, value: float, alert: bool, upper_threshold: Optional[float] = None, lower_threshold: Optional[float] = None)[source]

Bases: Metric

Represents results of the DLE estimator.

Stored in the dle_performance_metrics table.

Create a new model by parsing and validating input data from keyword arguments.

Raises ValidationError if the input data cannot be parsed to form a valid model.

alert: bool: Indicates if the method raised an alert for this Chunk

end_timestamp: datetime: The end datetime of the Chunk

id: Optional[int]: The technical identifier for this database row

lower_threshold: Optional[float]: The lower alerting threshold value

metric_name: str: The name of the method being calculated, e.g. jensen_shannon or chi2

model_id: Optional[int]: Foreign key pointing to a record in the model table

run_id: int: Foreign key pointing to a record in the run table

start_timestamp: datetime: The start datetime of the Chunk

timestamp: datetime: The ‘’center’’ timestamp of the Chunk, i.e. the mean of the start and end timestamps

upper_threshold: Optional[float]: The upper alerting threshold value

value: float: The value returned by the method

class nannyml.io.db.entities.DataReconstructionFeatureDriftMetric(*, id: Optional[int] = None, model_id: Optional[int] = None, run_id: int = None, start_timestamp: datetime, end_timestamp: datetime, timestamp: datetime, metric_name: str, value: float, alert: bool, upper_threshold: Optional[float] = None, lower_threshold: Optional[float] = None)[source]

Bases: Metric

DataReconstructionDriftCalculator results.

Stored in the data_reconstruction_feature_drift_metrics table.

Create a new model by parsing and validating input data from keyword arguments.

Raises ValidationError if the input data cannot be parsed to form a valid model.

alert: bool: Indicates if the method raised an alert for this Chunk

end_timestamp: datetime: The end datetime of the Chunk

id: Optional[int]: The technical identifier for this database row

lower_threshold: Optional[float]: The lower alerting threshold value

metric_name: str: The name of the method being calculated, e.g. jensen_shannon or chi2

model_id: Optional[int]: Foreign key pointing to a record in the model table

run_id: int: Foreign key pointing to a record in the run table

start_timestamp: datetime: The start datetime of the Chunk

timestamp: datetime: The ‘’center’’ timestamp of the Chunk, i.e. the mean of the start and end timestamps

upper_threshold: Optional[float]: The upper alerting threshold value

value: float: The value returned by the method

class nannyml.io.db.entities.Metric(*, id: Optional[int] = None, model_id: Optional[int] = None, run_id: int = None, start_timestamp: datetime, end_timestamp: datetime, timestamp: datetime, metric_name: str, value: float, alert: bool)[source]

Bases: SQLModel

Base Metric definition.

Create a new model by parsing and validating input data from keyword arguments.

Raises ValidationError if the input data cannot be parsed to form a valid model.

alert: bool: Indicates if the method raised an alert for this Chunk

end_timestamp: datetime: The end datetime of the Chunk

id: Optional[int]: The technical identifier for this database row

metric_name: str: The name of the method being calculated, e.g. jensen_shannon or chi2

model_id: Optional[int]: Foreign key pointing to a record in the model table

run_id: int: Foreign key pointing to a record in the run table

start_timestamp: datetime: The start datetime of the Chunk

timestamp: datetime: The ‘’center’’ timestamp of the Chunk, i.e. the mean of the start and end timestamps

value: float: The value returned by the method

class nannyml.io.db.entities.MissingValuesMetric(*, id: Optional[int] = None, model_id: Optional[int] = None, run_id: int = None, start_timestamp: datetime, end_timestamp: datetime, timestamp: datetime, metric_name: str, value: float, alert: bool, column_name: str, upper_threshold: Optional[float] = None, lower_threshold: Optional[float] = None)[source]

Bases: Metric

Create a new model by parsing and validating input data from keyword arguments.

Raises ValidationError if the input data cannot be parsed to form a valid model.

alert: bool: Indicates if the method raised an alert for this Chunk

column_name: str: The name of the column this metric belongs to

end_timestamp: datetime: The end datetime of the Chunk

id: Optional[int]: The technical identifier for this database row

lower_threshold: Optional[float]: The lower alerting threshold value

metric_name: str: The name of the method being calculated, e.g. jensen_shannon or chi2

model_id: Optional[int]: Foreign key pointing to a record in the model table

run_id: int: Foreign key pointing to a record in the run table

start_timestamp: datetime: The start datetime of the Chunk

timestamp: datetime: The ‘’center’’ timestamp of the Chunk, i.e. the mean of the start and end timestamps

upper_threshold: Optional[float]: The upper alerting threshold value

value: float: The value returned by the method

class nannyml.io.db.entities.Model(*, id: Optional[int] = None, name: str)[source]

Bases: SQLModel

Represents a Model.

Only created when the model_name property of the DatabaseWriter was given. The id field here will act as a foreign key in the run table and all metric tables.

Stored in the model table.

Create a new model by parsing and validating input data from keyword arguments.

Raises ValidationError if the input data cannot be parsed to form a valid model.

id: Optional[int]: A technical key that is used as a foreign key in the other tables

name: str: Optional model name that might be useful in visualizations e.g. in Grafana dashboards

runs: List[Run]: List of NannyML runs

class nannyml.io.db.entities.RealizedPerformanceMetric(*, id: Optional[int] = None, model_id: Optional[int] = None, run_id: int = None, start_timestamp: datetime, end_timestamp: datetime, timestamp: datetime, metric_name: str, value: float, alert: bool, upper_threshold: Optional[float] = None, lower_threshold: Optional[float] = None)[source]

Bases: Metric

Represents results of the PerformanceCalculator.

Stored in the realized_performance_metrics table.

Create a new model by parsing and validating input data from keyword arguments.

Raises ValidationError if the input data cannot be parsed to form a valid model.

alert: bool: Indicates if the method raised an alert for this Chunk

end_timestamp: datetime: The end datetime of the Chunk

id: Optional[int]: The technical identifier for this database row

lower_threshold: Optional[float]: The lower alerting threshold value

metric_name: str: The name of the method being calculated, e.g. jensen_shannon or chi2

model_id: Optional[int]: Foreign key pointing to a record in the model table

run_id: int: Foreign key pointing to a record in the run table

start_timestamp: datetime: The start datetime of the Chunk

timestamp: datetime: The ‘’center’’ timestamp of the Chunk, i.e. the mean of the start and end timestamps

upper_threshold: Optional[float]: The upper alerting threshold value

value: float: The value returned by the method

class nannyml.io.db.entities.Run(*, id: Optional[int] = None, model_id: Optional[int] = None, execution_timestamp: datetime = datetime.datetime(2024, 3, 8, 13, 18, 34, 984806))[source]

Bases: SQLModel

Represents a NannyML run, allowing to filter results based on what run generated them.

The id field here will act as a foreign key in all metric tables.

Stored in the run table.

Create a new model by parsing and validating input data from keyword arguments.

Raises ValidationError if the input data cannot be parsed to form a valid model.

execution_timestamp: datetime: Execution time of NannyML run

id: Optional[int]: Foreign key in all metric tables

model: Model: The actual Model class instance that is linked to the run

model_id: Optional[int]: Used to link a run to a model

class nannyml.io.db.entities.UnivariateDriftMetric(*, id: Optional[int] = None, model_id: Optional[int] = None, run_id: int = None, start_timestamp: datetime, end_timestamp: datetime, timestamp: datetime, metric_name: str, value: float, alert: bool, column_name: str)[source]

Bases: Metric

Represents results of the UnivariateDriftCalculator.

Stored in the univariate_drift_metrics table.

Create a new model by parsing and validating input data from keyword arguments.

Raises ValidationError if the input data cannot be parsed to form a valid model.

alert: bool: Indicates if the method raised an alert for this Chunk

column_name: str: The name of the column this metric belongs to

end_timestamp: datetime: The end datetime of the Chunk

id: Optional[int]: The technical identifier for this database row

metric_name: str: The name of the method being calculated, e.g. jensen_shannon or chi2

model_id: Optional[int]: Foreign key pointing to a record in the model table

run_id: int: Foreign key pointing to a record in the run table

start_timestamp: datetime: The start datetime of the Chunk

timestamp: datetime: The ‘’center’’ timestamp of the Chunk, i.e. the mean of the start and end timestamps

value: float: The value returned by the method

class nannyml.io.db.entities.UnseenValuesMetric(*, id: Optional[int] = None, model_id: Optional[int] = None, run_id: int = None, start_timestamp: datetime, end_timestamp: datetime, timestamp: datetime, metric_name: str, value: float, alert: bool, column_name: str, upper_threshold: Optional[float] = None, lower_threshold: Optional[float] = None)[source]

Bases: Metric

Create a new model by parsing and validating input data from keyword arguments.

Raises ValidationError if the input data cannot be parsed to form a valid model.

alert: bool: Indicates if the method raised an alert for this Chunk

column_name: str: The name of the column this metric belongs to

end_timestamp: datetime: The end datetime of the Chunk

id: Optional[int]: The technical identifier for this database row

lower_threshold: Optional[float]: The lower alerting threshold value

metric_name: str: The name of the method being calculated, e.g. jensen_shannon or chi2

model_id: Optional[int]: Foreign key pointing to a record in the model table

run_id: int: Foreign key pointing to a record in the run table

start_timestamp: datetime: The start datetime of the Chunk

timestamp: datetime: The ‘’center’’ timestamp of the Chunk, i.e. the mean of the start and end timestamps

upper_threshold: Optional[float]: The upper alerting threshold value

value: float: The value returned by the method