chemicalchecker.tool.targetmate.evaluation.validation.Validation

class Validation(splitter=None, is_cv=False, is_stratified=True, n_splits=3, test_size=0.2, explain=False, model_type=None, only_train=False, only_validation=True, **kwargs)

Bases: BaseValidation, HPCUtils

Initialize validation class.

Parameters:

• splitter (object) – If none specified, the corresponding TargetMate splitter is used.

• is_cv (bool) – If False, a simple train-test split is done.

• is_stratified (bool) – Do stratified split.

• n_splits (int) – Number of splits to perform.

• explain (bool) – Calculate Shapley values for predictions.

• test_size (float) – Proportion of samples in the test set.

• is_tmp (bool) – Store all in temporary path

• is_tmp_bases|signatures|predictions (bool) – Store bases|signatures|predictions in temporary path

Methods

as_dict
cpu_count
fit
func_hpc	Execute the any method on the configured HPC.
gather
get_splits
multi_validate
multi_validate_onlytrain
predict
save
score
setup
single_validate
validate	Validate a TargetMate model using train-test splits.
waiter	Wait for jobs to finish

func_hpc(func_name, *args, **kwargs)

Execute the any method on the configured HPC.

Parameters:

• args (tuple) – the arguments for of the function method

• kwargs (dict) – arguments for the HPC method.

validate(tm, data, train_idx=None, test_idx=None, wipe=True, scramble=False, set_train_idx=None, set_test_idx=None)

Validate a TargetMate model using train-test splits.

Parameters:

• tm (TargetMate model, or list of) – The TargetMate model to be evaluated. A list is accepted.

• data (InputData, or list of) – Data object. A list is accepted.

• train_idx (array) – Precomputed indices for the train set (default=None).

• test_idx (array) – Precomputed indices for the test set (default=None).

• wipe (bool) – Clean temporary directory once done (default=True).

• only_validation (bool) – Only the validations files pkl and txt files are kept (default=True).

waiter(jobs, secs=3)

Wait for jobs to finish