DRecPy.Evaluation package

• DRecPy.Evaluation.Metrics package
  • DRecPy.Evaluation.Metrics.ranking module
  • DRecPy.Evaluation.Metrics.regression module
• DRecPy.Evaluation.Splits package
  • DRecPy.Evaluation.Splits.leave_k_out module
  • DRecPy.Evaluation.Splits.matrix_split module
  • DRecPy.Evaluation.Splits.random_split module
• DRecPy.Evaluation.Processes package
  • DRecPy.Evaluation.Processes.predictive_evaluation module
  • DRecPy.Evaluation.Processes.ranking_evaluation module
  • DRecPy.Evaluation.Processes.recommendation_evaluation module

DRecPy.Evaluation.loss_tracker module

class DRecPy.Evaluation.loss_tracker.LossTracker

Bases: object

add_epoch_callback_result(name, result, epoch)

Adds a new epoch callback result.

Parameters:

• name – A string representing the name of the evaluated metric.
• result – A number representing the evaluated value for the current metric.
• epoch – A number representing the epoch for which the evaluated metric showed the passed result.

add_epoch_loss(loss)

Adds a new epoch loss.

Parameters:loss – The loss value obtained during the epoch.

display_graph(model_name=None, stopping_epoch=None, block=False)

Displays a graph containing the average batch loss per epoch, as well as the epoch callback results for each metric, if they exist.

Parameters:

• model_name – A string representing the name of the model. If this is provided, the model name will be displayed on the title of the figure. Default: None.
• stopping_epoch – An integer representing the epoch for which early stopping has been applied and therefore all network weights were reverted to.
• block – A boolean indicating whether the displayed graph should block code execution or not. Default: False.

get_epoch_avg_loss()

Gets the current average epoch loss.

Returns:The current average epoch loss, computed from the provided epoch losses.

reset_epoch_losses()

Resets the stored epoch losses and sets the epoch average loss to 0.