Base Class

Base class for classical methods implementation.

This section contains the abstract base class that defines the interface for all classical machine learning methods in TALENT. The base class provides common functionality for data preprocessing, model training, and evaluation.

TALENT.model.classical_methods.base.check_softmax(logits)
class TALENT.model.classical_methods.base.classical_methods(args, is_regression)

Bases: object

construct_model(model_config=None)
data_format(is_train=True, N=None, C=None, y=None)
fit(data, info, train=True, config=None)
metric(predictions, labels, y_info)
reset_stats_withconfig(config)
class TALENT.model.classical_methods.base.classical_methods

Abstract base class for all classical machine learning methods in TALENT.

Key Features:

  • Handles common data preprocessing including missing value imputation, categorical encoding, numerical encoding and binning, data normalization, and label processing

  • Provides unified evaluation metrics calculation

  • Supports both regression and classification tasks

  • Automatically handles data format conversion between training and testing

  • Manages model saving and loading functionality

Abstract Methods:

  • construct_model(): Must be implemented by subclasses to create the specific model

  • fit(): Must be implemented by subclasses for model training

  • predict(): Must be implemented by subclasses for model prediction

__init__(args, is_regression)

Initialize the classical method.

Parameters:

  • args (object) – Configuration arguments containing model settings

  • is_regression (bool) – Whether the task is regression (True) or classification (False)

data_format(is_train=True, N=None, C=None, y=None)

Format data for training or testing.

Parameters:

  • is_train (bool, default=True) – Whether processing training data or test data

  • N (array-like, optional) – Numerical features data

  • C (array-like, optional) – Categorical features data

  • y (array-like, optional) – Target labels

construct_model(model_config=None)

Construct the specific model instance.

Parameters:

  • model_config (dict, optional) – Model configuration parameters

Abstract Method: Must be implemented by subclasses.

fit(data, info, train=True, config=None)

Train the model on the provided data.

Parameters:

  • data (tuple) – Tuple containing (N, C, y) where N is numerical features, C is categorical features, y is labels

  • info (dict) – Dataset information

  • train (bool, default=True) – Whether to train the model or just load from checkpoint

  • config (dict, optional) – Additional configuration parameters

Returns:

  • time_cost (float) – Training time in seconds

Abstract Method: Must be implemented by subclasses.

reset_stats_withconfig(config)

Reset statistics with new configuration.

Parameters:

  • config (dict) – Configuration parameters

metric(predictions, labels, y_info)

Calculate evaluation metrics.

Parameters:

  • predictions (array-like) – Model predictions

  • labels (array-like) – True labels

  • y_info (dict) – Label information

Returns:

  • vres (tuple) – Evaluation metrics values

  • metric_name (tuple) – Names of the evaluation metrics

Metrics:

  • For regression: returns MAE, R2, RMSE metrics

  • For classification: returns Accuracy, Avg_Precision, Avg_Recall, F1 metrics

References:

[1] TALENT Framework Documentation. Classical Methods Base Class.