Event_Model Class

This class provides a framework for modeling soccer events, including training and inference capabilities using various model architectures.

Initialization

__init__(model_name, config=None)

Initializes the event model with the specified model name and configuration.

Parameters:

• model_name –

  (str) Name of the model to be used. Supported models include:

  • ’FMS’

  • ’LEM_action’

  • ’LEM’

  • ’MAJ’

  • ’NMSTPP’

  • ’Seq2Event’

• config – (str, optional) Path to a YAML configuration file. If not provided, the default settings will be loaded.

Raises:

• FileNotFoundError – If the specified configuration file does not exist.

• ValueError – If the model name is ‘MAJ’, Optuna will be disabled.

Methods

train()

Trains the specified model based on the configuration. The method selects the appropriate training routine depending on the model name and whether Optuna is enabled.

Raises:

ValueError – If the model name is unknown.

inference(model_path, model_config, train_path=None, valid_path=None, save_path=None, simulation=False, random_selection=True, max_iter=20, min_max_dict_path=None)

Conducts inference using the trained model. It saves the results to specified paths and can perform simulations if indicated.

Parameters:

• model_path – (str) Path to the saved model.

• model_config – (str) Path to the JSON configuration file for the model.

• train_path – (str, optional) Path to the training dataset. Defaults to the path defined in the model configuration if not provided.

• valid_path – (str, optional) Path to the validation dataset. Defaults to the path defined in the model configuration if not provided.

• save_path – (str, optional) Directory to save inference results. Defaults to the path defined in the model configuration.

• simulation – (bool, optional) Flag indicating whether to perform a simulation. Defaults to False.

• random_selection – (bool, optional) Flag for random selection in simulation. Defaults to True.

• max_iter – (int, optional) Maximum number of iterations for simulation. Defaults to 20.

• min_max_dict_path – (str, optional) Path to the min-max normalization dictionary.

Raises:

ValueError – If neither train_path nor min_max_dict_path are provided, or if the model name is unknown.

Returns:

Depending on the simulation flag, returns either the inferenced data or simulation results along with evaluation metrics.