Split Dataset (Experimental)

Splits input dataset into a training and a test dataset in a lazy-loaded manner. This means it only adds the indices attribute in the dataset structure, indicating the indices of the dataset that belong to the Training or Test datasets without actually splitting the dataset into two. As a result, you will not see the actual splits in the node’s output ports.

Definition

Input ports

input_ds dataset

Input dataset

Output ports

train_ds dataset

Training dataset

test_ds dataset

Test dataset

Configuration

Create image labels from filepaths (img_path). (labels)

Specify expression to create label column for image datasets or column name for tabular datasets

Choose an output column (output_col)

Label column

Stratify (stratify)

Stratify data using Y as class labels

Test set proportion (test_size)

Test size for train/test split

Implementation

class node_splitdataset.SplitDataset