mindmeld.active_learning.alp module

This module contains the ActiveLearningPipeline class.

class mindmeld.active_learning.alp.ActiveLearningPipeline(app_path: str, train_pattern: str, test_pattern: str, train_seed_pct: float, n_classifiers: int, n_epochs: int, batch_size: int, classifier_tuning_strategies: list, tagger_tuning_strategies: list, tuning_level: list, classifier_selection_strategy: str, tagger_selection_strategy: str, save_sampled_queries: bool, aggregate_statistic: str, class_level_statistic: str, log_usage_pct: float, labeled_logs_pattern: str, unlabeled_logs_path: str, output_folder: str)

Bases: object

Class that executes the strategy tuning and query selection process for the Active Learning Pipeline.

plot()

Creates the generated folder and its subfolders if they do not already exist.

select_queries()

Selects the next batch of queries to label from a set of log queries.

tune_strategies()

Loads the initial data bucket and then trains on every strategy.

num_iterations

Calculates the number of iterations needed for tuning. :returns: Number of iterations needed for tuning. :rtype: num_iterations (int)

class mindmeld.active_learning.alp.ActiveLearningPipelineFactory

Bases: object

Creates an ActiveLearningPipeline instance from values in a config.

static create_from_config(config)

Creates an augmentor instance using the provided configuration :param config: A model configuration. :type config: dict

Returns:An ActiveLearningPipeline class Return type:ActiveLearningPipeline Raises:ValueError -- When model configuration is invalid or required key is missing