Back to models
Transformer

Logger

API referenceSource
MultivariateInverse transformMissing valuesUnequal-length series

Logging transformer, writes data to logging, and otherwise leaves it unchanged.

In methods, logs X and y to logger. The logger can us as logger_backend a python logging instance, primarily for printing, with data logged as extra, or a custom DataLog instance to retrieve full objects and not just printouts.

Quickstart

python
from sktime.transformations.compose import Logger

estimator = Logger(logger='sktime', logger_backend='logging', log_methods='all', level=None, log_fitted_params=False)

Tags

Capabilities

Properties

Input typescitype:transform-input
Series
Output typescitype:transform-output
Series
Label typescitype:transform-labels
None
Fit is emptyfit_is_empty
No
Keeps the time indextransform-returns-same-time-index
Yes
Requires Xrequires_X
Yes
Requires yrequires_y
No
X and y need the same indexX-y-must-have-same-index
No

Parameters(5)

loggerstr optional, default=”sktime”

logger name to use, passed to logger_backend to identify the unique logger instance referenced by get_logger.

logger_backendstr, one of “logging” (default), “datalog”

Backend to use for logging.

  • “logging”: uses the standard Python logging module, logs to logging.getLogger(logger)

  • “datalog”: uses a multiton logger class for easy retrieval of data, logs to DataLog(logger), with DataLog from the transformations.compose module.

In either case, a reference to the logger can be retrieved by calling obj.get_logger, where obj is an instance of Logger.

log_methodsstr or list of str, default=``”transform”``

if "all", will log fit, transform, inverse_transform; if str or list of str, all strings must be from among the above, and will log exactly the methods that are passed as str; can also be "off"" to disable logging entirely.

levellogging level, optional, default=logging.INFO

logging level, one of logging.INFO, logging.DEBUG, logging.WARNING, logging.ERROR

log_fitted_paramsbool, optional, default=False

if True, will also write X and y seen in fit to self as X_ and y_, these can be retrieved by calling get_fitted_params. If False, get_fitted_params will return an empty dict.

Examples

>>> from sktime.transformations.compose import DataLog, Logger
>>> from sktime.datasets import load_airline
>>> from sktime.forecasting.naive import NaiveForecaster
>>> from sktime.transformations.detrend import Detrender
>>> 
>>> # create a logger
>>> logger = Logger (logger = "foo", log_methods = "all", logger_backend = "datalog")
>>> # create a pipeline that logs after detrending and before forecasting
>>> pipe = Detrender () * logger * NaiveForecaster (sp = 12)
>>> pipe. fit (load_airline (), fh = [1, 2, 3 ]) TransformedTargetForecaster(
... )
>>> # get the log
>>> log = DataLog ("foo"). get_log ()