def test_cols_set():
"""
Tests if the function correctly throws an error if cols is a set
"""
with pytest.raises(AssertionError, match="Cols must be a list!"):
pyedahelper.fast_missing_impute(df=test_data,
method="mean",
cols={'R', 'Python'})
def test_mode_fail():
"""
Tests if the function correctly throws an error if method is mode
and one or more of the columns don't have a mode
"""
with pytest.raises(Exception):
pyedahelper.fast_missing_impute(df=test_data,
method="mode",
cols=["c"])
def test_method_format():
"""
Tests if the function correctly throws an error if method is not a valid
method
"""
with pytest.raises(AssertionError, match="Not a valid method!"):
pyedahelper.fast_missing_impute(df=test_data,
method="avg",
cols=["a", "b"])
def test_one_method():
"""
Tests if the function correctly throws an error if the method is not a
string when more than one method is inputted
"""
with pytest.raises(AssertionError, match="Method must be a string!"):
pyedahelper.fast_missing_impute(df=test_data,
method=["mean", "median"],
cols="a")
def test_cols_allstr():
"""
Tests if the function correctly throws an error if cols is a list that
contains a non-string element
"""
with pytest.raises(AssertionError,
match="Columns must be a list of strings"):
pyedahelper.fast_missing_impute(df=test_data,
method="median",
cols=["a", 3, "c"])
def test_non_numeric():
"""
Tests if the function correctly throws an error if an column in cols is
non-numeric and the method selected is not 'remove' or 'mode'
"""
with pytest.raises(AssertionError,
match="With non-numeric columns, can only use "
"method = 'remove' or 'mode'"):
pyedahelper.fast_missing_impute(df=test_data,
method="mean",
cols=["e"])
def test_cols_in_df():
"""
Tests if the function correctly throws an error if an element in cols
is not part of the data frame
"""
with pytest.raises(AssertionError,
match="One or more of the column names"
" are not in the data frame!"):
pyedahelper.fast_missing_impute(df=test_data,
method="mean",
cols=["a", "c", "fake_col"])
def test_remove_method():
"""
Tests if the function, when method = 'remove', leaves the selected
column(s) with no NAs and leaves the other columns the same
"""
sample_drop = pyedahelper.fast_missing_impute(df=test_data,
method="remove",
cols=["a", "e"])
assert sample_drop["a"].isna().sum() == 0
assert sample_drop["e"].isna().sum() == 0
assert sample_drop["f"].isna().sum() == 4
assert sample_drop.shape[0] < test_data.shape[
0] # should have fewer rows because NAs are removed
assert sample_drop.shape[1] == test_data.shape[
1] # number of columns should not change
def test_mode():
"""
Tests if the function, when method = 'mode', leaves the selected columns
with NAs imputed with the mode of the column, with no NAs left in the
column, and leaves the other columns the same
"""
sample_mode = pyedahelper.fast_missing_impute(df=test_data,
method="mode",
cols=["a", "e"])
assert sample_mode["a"].isna().sum() == 0
assert sample_mode["e"].isna().sum() == 0
assert sample_mode["d"].isna().sum() == test_data["d"].isna().sum(
) # should remain unchanged
assert sample_mode["a"][5] == mode(test_data["a"])
assert sample_mode["e"][1] == mode(test_data["e"])
assert sample_mode.shape == test_data.shape # dimensions shouldn't change
def test_mean():
"""
Tests if the function, when method = 'mean', leaves the selected columns
with NAs imputed with the mean of the column, with no NAs left in the
column, and leaves the other columns the same
"""
sample_mean = pyedahelper.fast_missing_impute(df=test_data,
method="mean",
cols=["a", "c"])
assert sample_mean["a"].isna().sum() == 0
assert sample_mean["c"].isna().sum() == 0
assert sample_mean["b"].isna().sum() == test_data["b"].isna().sum(
) # should remain unchanged
assert sample_mean["a"][5] == test_data["a"].mean()
assert sample_mean["c"][2] == test_data["c"].mean()
assert sample_mean.shape == test_data.shape # dimensions should not change
def test_df_assert():
"""
Tests if the function correctly throws an error if df is not a dataframe
"""
with pytest.raises(AssertionError, match="Data must be a data frame!"):
pyedahelper.fast_missing_impute(df=test_list, method="mean", cols="a")
