immuneML.ml_methods.util package

Submodules

immuneML.ml_methods.util.Util module

class immuneML.ml_methods.util.Util.Util

Bases: object

static binarize_label_classes(true_y, predicted_y, classes)

Binarizes the predictions in place using scikit-learn’s label_binarize() method

Necessary for some sklearn metrics, like roc_auc_score

static get_immuneML_version()

static get_train_val_indices(n_examples, training_percentage, random_seed=None)

static make_binary_class_mapping(y, label: Label) → dict

Creates binary class mapping from a list of classes which can be strings, numbers or boolean values

Parameters:

• y – list of classes per example, as supplied to fit() method of the classifier; it should include all

• data (classes that will appear in the)

• label – label object along with information on the positive label

Returns:

mapping dictionary where 0 and 1 are always the keys and the values are original class names which were mapped for these values

static make_class_mapping(y, label: Label) → dict

Creates a class mapping from a list of classes which can be strings, numbers of booleans; maps to same name in multi-class settings

static map_to_new_class_values(y, class_mapping: dict)

static map_to_old_class_values(y, class_mapping: dict)

static setup_pytorch(number_of_threads, random_seed, pytorch_device_name=None)

static subset_encoded_data(encoded_data: EncodedData, indices)

immuneML.ml_methods.util.pytorch_util module

immuneML.ml_methods.util.pytorch_util.store_weights(model, path)

Module contents