def test_wrong_range():
small_list = np.random.normal(5, 2, 3)
large_list = np.random.normal(5, 2, 6000)
with pytest.raises(ValueError):
shapiro_wilk(small_list)
with pytest.raises(ValueError):
shapiro_wilk(large_list)
def test_meaningless_input():
"""check that meaningless input isn't prcoessed"""
with pytest.raises(ValueError):
shapiro_wilk(data_meaningless1)
with pytest.raises(ValueError):
shapiro_wilk(data_meaningless2)
with pytest.raises(TypeError):
shapiro_wilk(data_meaningless3)
with pytest.raises(ValueError):
shapiro_wilk(data_meaningless4)
with pytest.raises(ValueError):
shapiro_wilk(data_meaningless5)
def test_output_and_input_type():
"""Check that output is a tuple for all possible inputs"""
assert type(shapiro_wilk(data_df)) == tuple
assert type(shapiro_wilk(data_list1)) == tuple
assert type(shapiro_wilk(data_list2)) == tuple
assert type(shapiro_wilk(data_list3)) == tuple
assert type(shapiro_wilk(data_ndarray)) == tuple
assert type(shapiro_wilk(data_ndarray[:,0])) == tuple
assert type(shapiro_wilk(data_series1)) == tuple
assert type(shapiro_wilk(data_series2)) == tuple
def calculation_test():
"""check that the shapiro-wilk test statistic is correctly calculated because p-value should be > 0.05"""
norm_values = np.random.normal(5, 2, 100)
assert shapiro_wilk(norm_values)[1] > 0.05
def test_input_type():
"""check that the function returns TypeError if the dataframe has no continuous variables"""
with pytest.raises(TypeError):
shapiro_wilk(data_df2)
def test_output_lists_equal():
"""check that the length of the first list is equal to the output of the second list"""
stats = shapiro_wilk(data)
assert len(stats[0]) == len(stats[1])
def test_output_list_length():
"""check that the length of the first list is as <= amount of columns in pd.dataframe"""
stats_1 = shapiro_wilk(data_df)[0]
assert len(stats_1) <= data_df.shape[1]
def test_output_tuple_length():
"""check that length of output is two since the tuple always has 2 lists in it"""
stats = shapiro_wilk(data_df)
assert len(stats) == 2
def test_output_type():
"""Check that output is a tuple"""
stats = shapiro_wilk(data_df)
assert type(stats) == tuple