def test_visualize_feature_remove_na():
loan_data_dup = pandas.DataFrame()
loan_data_dup["emp_length_int"] = loan_data["emp_length_int"]
loan_data_dup = pandas.concat([
loan_data_dup,
pandas.DataFrame([numpy.nan] * 250, columns=["emp_length_int"])
],
ignore_index=True).sample(frac=1,
random_state=0)
assert loan_data_dup["emp_length_int"].isna().sum() == 250
visualize_feature(loan_data_dup["emp_length_int"], remove_na=True)
result_path = Path(__file__).parents[0].absolute().joinpath(
"result_images").joinpath("test_visualization_aids").joinpath(
"test_visualize_feature_remove_na.png")
pyplot.savefig(str(result_path))
baseline_path = Path(__file__).parents[0].absolute().joinpath(
"baseline_images").joinpath("test_visualization_aids").joinpath(
"test_visualize_feature_remove_na.png")
pyplot.cla()
pyplot.close(pyplot.gcf())
compare_images_from_paths(str(baseline_path), str(result_path))
def test_visualize_feature_int():
visualize_feature(loan_data["loan_condition_cat"])
result_path = Path(__file__).parents[0].absolute().joinpath(
"result_images").joinpath("test_visualization_aids").joinpath(
"test_visualize_feature_int.png")
pyplot.savefig(str(result_path))
baseline_path = Path(__file__).parents[0].absolute().joinpath(
"baseline_images").joinpath("test_visualization_aids").joinpath(
"test_visualize_feature_int.png")
pyplot.cla()
pyplot.close(pyplot.gcf())
compare_images_from_paths(str(baseline_path), str(result_path))
def test_visualize_feature_object():
visualize_feature(loan_data["income_category"])
result_path = Path(__file__).parents[0].absolute().joinpath(
"result_images").joinpath("test_visualization_aids").joinpath(
"test_visualize_feature_object.png")
pyplot.gcf().set_size_inches(10, 8)
pyplot.savefig(str(result_path))
baseline_path = Path(__file__).parents[0].absolute().joinpath(
"baseline_images").joinpath("test_visualization_aids").joinpath(
"test_visualize_feature_object.png")
pyplot.cla()
pyplot.close(pyplot.gcf())
compare_images_from_paths(str(baseline_path), str(result_path))
def test_visualize_feature_category_more_than_10_categories():
visualize_feature(loan_data["purpose"].astype("category"))
result_path = Path(__file__).parents[0].absolute().joinpath(
"result_images").joinpath("test_visualization_aids").joinpath(
"test_visualize_feature_category_more_than_10_categories.png")
pyplot.gcf().set_size_inches(11, 11)
pyplot.savefig(str(result_path))
baseline_path = Path(__file__).parents[0].absolute().joinpath(
"baseline_images").joinpath("test_visualization_aids").joinpath(
"test_visualize_feature_category_more_than_10_categories.png")
pyplot.cla()
pyplot.close(pyplot.gcf())
compare_images_from_paths(str(baseline_path), str(result_path))
def test_visualize_feature_bool():
loan_dup = pandas.DataFrame()
loan_dup["term 36 months"] = loan_data["term"].apply(
lambda term: True if term == " 36 months" else False).astype("bool")
visualize_feature(loan_dup["term 36 months"])
result_path = Path(__file__).parents[0].absolute().joinpath(
"result_images").joinpath("test_visualization_aids").joinpath(
"test_visualize_feature_bool.png")
pyplot.savefig(str(result_path))
baseline_path = Path(__file__).parents[0].absolute().joinpath(
"baseline_images").joinpath("test_visualization_aids").joinpath(
"test_visualize_feature_bool.png")
pyplot.cla()
pyplot.close(pyplot.gcf())
compare_images_from_paths(str(baseline_path), str(result_path))
def test_visualize_feature_float_exist_ax():
pyplot.figure()
ax = pyplot.gca()
ax.set_title("My ax")
visualize_feature(loan_data["emp_length_int"], ax=ax)
result_path = Path(__file__).parents[0].absolute().joinpath(
"result_images").joinpath("test_visualization_aids").joinpath(
"test_visualize_feature_float_exist_ax.png")
pyplot.gcf().set_size_inches(10, 8)
pyplot.savefig(str(result_path))
baseline_path = Path(__file__).parents[0].absolute().joinpath(
"baseline_images").joinpath("test_visualization_aids").joinpath(
"test_visualize_feature_float_exist_ax.png")
pyplot.cla()
pyplot.close(pyplot.gcf())
compare_images_from_paths(str(baseline_path), str(result_path))