search

shivashankarrs / classimb_fairness

5 stars 0 forks source link

Task 9 (experiments API design) #9

Open shivashankarrs opened 3 years ago

shivashankarrs commented 3 years ago

This design can help guide how to create all other dependent codes, e.g.,

data = get_dataset("dataset_name") model = class_imbalance_learner(train = dataset.train, val = dataset.val, approach = 1) //we can try various approaches metrics = model.evaluate(dataset.test) // evaluate for both fairness and class imbalance

With this we can loop over several datasets and approaches.

shivashankarrs commented 3 years ago

random seeds, class imbalance ratios (if we simulate), dataset should have private attributes and the column/key (for both private attributes and main class) has to be mentioned for evaluation.