def test_non_accepted_cat():
try:
fill_missing(X_train,
X_test,
col_dict,
num_imp="median",
cat_imp="mean")
except AssertionError:
pass
def test_bad_col_dict():
try:
fill_missing(X_train,
X_test,
column_dict=["cat1", "num1"],
num_imp="median",
cat_imp="mode")
except AssertionError:
pass
def test_bad_dict_keys():
try:
fill_missing(
X_train,
X_test,
# insert wrong column key
column_dict={
'numerical': ['num1'],
'categorical': ['cat1']
},
num_imp="median",
cat_imp="mean")
except AssertionError:
pass
def test_no_name_columns():
try:
fill_missing(
X_train=pd.DataFrame([1.5, 2.5, 3.5, None, 4.5]),
X_test=pd.DataFrame([1, None, 3, 1, 3]),
# insert column which is not in the df
column_dict={
'numeric': ['num2'],
'categorical': ['cat1']
},
num_imp="median",
cat_imp="mean")
except AssertionError:
pass
def test_bad_test_set():
try:
fill_missing(X_train=pd.DataFrame({
'cat1': [1, None, 3, 1, 3],
'num1': [1.5, 2.5, None, 2.0, 2.0]
}),
X_test=np.array(1),
column_dict={
'numeric': ['num1'],
'categorical': ['cat1']
},
num_imp="median",
cat_imp="mode")
except AssertionError:
pass
def run_pylaundry():
"""
Runs all modules of pylaundry
Arguments
--------
NA
Returns
------
features_selected = list of final features selected
"""
col_dict = categorize(df=X_train)
# second function - fill_missing
clean_data = fill_missing(X_train,
X_test,
col_dict,
num_imp="mean",
cat_imp="mode")
# third function - transform_columns
transformed_data = transform_columns(clean_data['X_train'],
clean_data['X_test'], col_dict)
# fourth function - feature selection
features_selected = select_features(transformed_data['X_train'],
y_train,
n_features=2)
return features_selected
def test_diff_columns():
try:
fill_missing(X_train=pd.DataFrame({
'cat1': [1, 2, None, 1, 1],
'num2': [1.5, 2.5, 3.5, None, 4.5]
}),
X_test=pd.DataFrame({
'cat1': [1, None, 3, 1, 3],
'num1': [1.5, 2.5, None, 2.0, 2.0]
}),
column_dict={
'numeric': ['num1'],
'categorical': ['cat1']
},
num_imp="median",
cat_imp="mean")
except AssertionError:
pass
def test_mode_cat():
output = fill_missing(X_train,
X_test,
col_dict,
num_imp="median",
cat_imp="mode")
train_output = output['X_train']
test_output = output['X_test']
# Mean column 1 is 3 for imputed value
# Check for same imputation train and test
assert train_output["cat1"][2] == 1, \
"Imputed mode value should be 1 in train set"
assert test_output["cat1"][1] == 1, \
"Imputed mode value should be 1 in test set"
def test_mean_num():
output = fill_missing(X_train,
X_test,
col_dict,
num_imp="mean",
cat_imp="mode")
train_output = output['X_train']
test_output = output['X_test']
# Mean column 1 is 3 for imputed value
# Check for same imputation train and test
assert train_output["num1"][3] == 3, \
"Imputed mean value should be 3.0 in train set"
assert test_output["num1"][2] == 3, \
"Imputed mean value should be 3.0 in test set"
def test_output_type():
output = fill_missing(X_train,
X_test,
col_dict,
num_imp="mean",
cat_imp="mode")
train_output = output['X_train']
test_output = output['X_test']
# Check length and type of output
assert len(output) == 2, \
"Output of fill_missing() should be two dataframes"
assert isinstance(train_output, pd.DataFrame), \
"Training df should be a Pandas DF"
assert isinstance(test_output, pd.DataFrame), \
"Test df should be a Pandas DF"
def test_non_numeric_columns():
output = fill_missing(X_train=pd.DataFrame({
'cat1': ['a', 'b', None, 'c', 'a'],
'num1': [1.5, 2.5, 3.5, None, 4.5]
}),
X_test=pd.DataFrame({
'cat1': ['a', 'b', None],
'num1': [1.5, None, 3.5],
}),
column_dict=col_dict,
num_imp="mean",
cat_imp="mode")
train_output = output['X_train']
test_output = output['X_test']
# Mode of column 1 is 3 for imputed value
# Check for same imputation train and test
assert train_output["cat1"][2] == 'a', \
"Imputed mode value should be 'a' in train set"
assert test_output["cat1"][2] == 'a', \
"Imputed mode value should be 'a' in test set"
def test_median_num():
output = fill_missing(X_train=pd.DataFrame({
'cat1': [1, 2, None, 1, 1],
'num1': [1.5, 2.5, 3.5, None, 4.5]
}),
X_test=pd.DataFrame({
'cat1': [1, None, 3, 1, 3],
'num1': [1.5, 2.5, None, 2.0, 2.0]
}),
column_dict={
'numeric': ['num1'],
'categorical': ['cat1']
},
num_imp="median",
cat_imp="mode")
train_output = output['X_train']
test_output = output['X_test']
# Mean column 1 is 3 for imputed value
# Check for same imputation train and test
assert train_output["num1"][3] == 3, \
"Imputed median value should be 3.0 in train set"
assert test_output["num1"][2] == 3, \
"Imputed median value should be 3.0 in test set"