def test_task5(test_data, dataset):
dataset_obj = test_data[dataset]["dataset_object"]
best_model = choose_best_model(dataset_obj)
if best_model is None:
pytest.fail("When testing Task #5, choose_best_model returned None instead of a Model object")
testing_R2 = validate_model(dataset_obj, best_model)
exp_training_R2 = test_data[dataset]["expected"]["validate_model"]["training_R2"]
exp_testing_R2 = test_data[dataset]["expected"]["validate_model"]["testing_R2"]
assert best_model.R2 == pytest.approx(exp_training_R2), "Incorrect Training R2"
assert testing_R2 == pytest.approx(exp_testing_R2), "Incorrect Testing R2"
def go(dataset, gen_html):
'''
Put together the work for all the tasks
Inputs: the dataset
'''
models = compute_single_var_models(dataset)
format_list_of_models("1a", models, gen_html=gen_html)
model = compute_all_vars_model(dataset)
format_list_of_models("1b", [model], gen_html=gen_html)
best_bivariate_model = compute_best_pair(dataset)
format_list_of_models("2", [best_bivariate_model], gen_html=gen_html)
models = backward_selection(dataset)
format_list_of_models("3", models, gen_html=gen_html)
best_model = choose_best_model(dataset)
format_list_of_models("4", [best_model],
include_adj_R2=True,
gen_html=gen_html)
if gen_html:
gh = ":"
tabs = ""
nl = " |br|\n"
else:
gh = ""
tabs = " "
nl = "\n"
s = gh + "Task 5{}\n".format(gh)
if best_model is None:
s += tabs + "Can't test this until Task 4 is implemented" + nl
else:
testing_R2 = validate_model(dataset, best_model)
s += tabs + str(best_model) + nl
if not hasattr(best_model, "R2"):
s += tabs + "Training R2: choose_best_model returned an object without an R2 model" + nl
else:
s += tabs + "Training R2: {}".format(best_model.R2) + nl
s += tabs + "Testing R2: {}".format(testing_R2) + nl
print(s)
def test_task4(test_data, dataset):
dataset_obj = test_data[dataset]["dataset_object"]
model = choose_best_model(dataset_obj)
check_models(test_data, dataset, "choose_best_model", [model])