DatasetGenerator

class hdnnpy.dataset.dataset_generator.DatasetGenerator(*datasets)

Bases: object

Deal out datasets as needed.

Parameters:*datasets (HDNNPDataset) – What you want to unite.

all()

Pass all datasets an instance have.

Returns:All stored datasets. Return type:list [HDNNPDataset]

foreach()

Pass all datasets an instance have one by one.

Returns:a stored dataset object. Return type:Iterator [HDNNPDataset]

holdout(ratio)

Split each dataset at a certain rate and pass it

Parameters:ratio (float) – Specify the rate you want to use as training data. Remains are test data. Returns:All stored dataset split by specified ratio into training and test data. Return type:list [tuple [HDNNPDataset, HDNNPDataset]]

kfold(kfold)

Split each dataset almost equally and pass it for cross validation.

Parameters:kfold (int) – Number of folds to split dataset. Returns:All stored dataset split into training and test data. It iterates k times while changing parts used for test data. Return type:Iterator [list [tuple [HDNNPDataset, HDNNPDataset]]]