autoresevaluator 0.1.7

No project description provided

AutoRes Evaluator

Architecture

How to use

• Install
  • https://pypi.org/project/autoresevaluator/

pip install autoresevaluator

from autoresevaluator import AutoResEvaluator

• Setting

# Hyperparameter setting
# Specify "type" and "args" for items to be searched in optuna.
params = {
    'lambda_l1': {'type': 'log_float', 'args': [1e-8, 10.0]},
    'lambda_l2': {'type': 'log_float', 'args': [1e-8, 10.0]},
    'num_leaves': {'type': 'int', 'args': [2, 256]},
    'feature_fraction': {'type': 'float', 'args': [0.4, 1.0]},
    'bagging_fraction': {'type': 'float', 'args': [0.4, 1.0]},
    'verbosity': -1
}


are = AutoResEvaluator(
    # task type
    task_type='tabledata binary classification',
    # dataset name
    dataset_name='titanic',
    # model file path
    model_path='/Users/tanakatouma/vscode/autores-evaluator/test/lightgbm_model.py',
    params=params,
    # Metrics you want to maximize/minimize
    valuation_index='pr_auc'
    )

• Execution

are.exec()

Output

• result.log

  • File to output the results

• model_error.log

  • File to write errors in model files