def __init__(self):
dataset = data.synthetic_item_response_theory()
del dataset['test_student_ids']
del dataset['test_question_ids']
del dataset['test_correct']
super(SyntheticItemResponseTheory,
self).__init__(name='synthetic_item_response_theory',
pretty_name='Synthetic Item-Response Theory',
**dataset)
def testSyntheticItemResponseTheory(self):
num_train_points = int(0.75 * 30012)
num_test_points = 30012 - num_train_points
dataset = data.synthetic_item_response_theory(train_fraction=0.75)
self.assertEqual((num_train_points,), dataset['train_student_ids'].shape)
self.assertEqual((num_train_points,), dataset['train_question_ids'].shape)
self.assertEqual((num_train_points,), dataset['train_correct'].shape)
self.assertEqual((num_test_points,), dataset['test_student_ids'].shape)
self.assertEqual((num_test_points,), dataset['test_question_ids'].shape)
self.assertEqual((num_test_points,), dataset['test_correct'].shape)
def synthetic_item_response_theory():
"""One-parameter logistic item-response theory (IRT) model.
This uses a dataset sampled from the prior. This dataset is a simulation of
400 students each answering a subset of 100 unique questions, with a total of
30012 questions answered.
Returns:
target: StanModel.
"""
dataset = data.synthetic_item_response_theory()
del dataset['test_student_ids']
del dataset['test_question_ids']
del dataset['test_correct']
return item_response_theory.item_response_theory(**dataset)