Module markov.api.recording.experiments.integrations.xgboost.xgboost_auto_record

Functions

def auto_record(name: str, notes: str = '', project_id: str = '', model_class: ModelClass = ModelClass.CLASSIFICATION, record_epoch_time: bool = True, **kwargs)

Automatically record your experiment with an xgboost model. This method needs to be called once before calling xgboost.train() or XGBModel.fit().

Args

name

Name of the experiment

notes

Notes for the experiment

project_id

Project id of the project where experiment should reside

model_class

type of model used for training - Classification, Regression etc. See markov.ModelClass

record_epoch_time

boolean - to switch recording "time taken per epoch"

**kwargs:

def get_callback(params)

def patch_xgboost_instance(wrapped, instance, args, kwargs, callback_params)

def patch_xgboost_kwargs(wrapped, args, kwargs, callback_params)

def patched_fit(wrapped, instance, args, kwargs)