def test_discretize_features_one_feature():
"""Test discretize_features for just one feature."""
recs = [{
'test_feature_1': 1,
'test_feature_2': 1,
'test_class': 1
}, {
'test_feature_1': 1,
'test_feature_2': 1,
'test_class': 1
}, {
'test_feature_1': 3,
'test_feature_2': 3,
'test_class': 1
}, {
'test_feature_1': 5,
'test_feature_2': 5,
'test_class': 0
}, {
'test_feature_1': 5,
'test_feature_2': 5,
'test_class': 0
}]
test_df = pd.DataFrame(recs)
obj = DataWrangling()
return_val = obj.discretize_features(data_frame=test_df,
features=['test_feature_1'],
n_bins=3)
expected_val = np.array([[1., 0., 0.], [1., 0., 0.], [0., 1., 0.],
[0., 0., 1.], [0., 0., 1.]])
assert np.array_equal(return_val, expected_val)
def test_fill_nul_vals():
"""Test fill_null_vals()."""
recs = [{
'test_feature_1': 0,
'test_feature_2': 0,
'test_class': 1
}, {
'test_feature_1': 1,
'test_feature_2': 5,
'test_class': 1
}, {
'test_feature_1': 2,
'test_feature_2': 0,
'test_class': 1
}, {
'test_feature_1': 1,
'test_feature_2': 3,
'test_class': 0
}, {
'test_feature_1': 0,
'test_feature_2': 2,
'test_class': 0
}]
test_df = pd.DataFrame(recs)
obj = DataWrangling()
return_val = obj.fill_nul_vals(
data_frame=test_df,
class_column='test_class',
features=['test_feature_1', 'test_feature_2'])
recs = [{
'test_feature_1': 1.5,
'test_feature_2': 5.0,
'test_class': 1
}, {
'test_feature_1': 1,
'test_feature_2': 5,
'test_class': 1
}, {
'test_feature_1': 2,
'test_feature_2': 5.0,
'test_class': 1
}, {
'test_feature_1': 1,
'test_feature_2': 3,
'test_class': 0
}, {
'test_feature_1': 1.0,
'test_feature_2': 2,
'test_class': 0
}]
expected_val = pd.DataFrame(recs)
assert return_val.equals(expected_val)
def test_split_train_test():
"""Test split_train_test()."""
recs = [{
'test_feature_1': 1,
'test_feature_2': 1,
'test_class': 1
}, {
'test_feature_1': 1,
'test_feature_2': 1,
'test_class': 1
}, {
'test_feature_1': 3,
'test_feature_2': 3,
'test_class': 1
}, {
'test_feature_1': 5,
'test_feature_2': 5,
'test_class': 0
}, {
'test_feature_1': 5,
'test_feature_2': 5,
'test_class': 0
}]
test_df = pd.DataFrame(recs)
features = ['test_feature_1', 'test_feature_2']
test_size = 0.33
class_col = 'test_class'
obj = DataWrangling()
return_val = obj.split_train_test(x_data=test_df[features].to_numpy(),
y_labels=test_df[class_col].to_numpy(),
features=features,
test_size=test_size)
expected_keys = ['x_train', 'y_train', 'x_test', 'y_test']
assert all(key in return_val for key in expected_keys)
assert all(
isinstance(val, (pd.Series, pd.DataFrame))
for _, val in return_val.items())
def test_sklearn_prediction_score():
global data, features, class_col, n_bins, test_size, no_zero_features
obj = DataWrangling()
splits = preprocess_data(obj=obj,
data=data,
features=features,
no_zero_features=no_zero_features,
class_col=class_col,
n_bins=n_bins,
test_size=test_size)
prediction_score = get_sklearn_prediction_score(splits)
splits = splits['x_train']
assert isinstance(prediction_score, float)
assert prediction_score >= 0.0
assert prediction_score <= 1.0
def test_preprocess_data():
global data, features, class_col, n_bins, test_size, no_zero_features
obj = DataWrangling()
return_val = preprocess_data(obj=obj,
data=data,
features=features,
no_zero_features=no_zero_features,
class_col=class_col,
n_bins=n_bins,
test_size=test_size)
expected_keys = ['x_train', 'y_train', 'x_test', 'y_test']
assert all(key in return_val for key in expected_keys)
assert all(
isinstance(val, (pd.Series, pd.DataFrame))
for _, val in return_val.items())
assert return_val['x_train'].shape[0] == 4
assert return_val['x_test'].shape[0] == 2
assert return_val['x_train'].shape[1] == 6
def get_data(n_bins: int) -> dict:
"""Retrieve data and perform preprocessing steps and return result.
Args:
n_bins (int): Number of bins to discretize features into.
Returns:
dict: Dictionary containing test and training sets and labels.
"""
dframe = pd.read_csv(TEST_DATA_PATH)
obj1 = DataWrangling(data_path=TEST_DATA_PATH)
obj1.load_data()
features = [
'Pglucose', 'Preg', 'Dbp', 'BMI', 'Dpfunc', 'Age', 'Insulin', 'Skin'
]
input_df = obj1.data[features + ['Response']].astype(np.float64)
no_zero_vals = obj1.fill_nul_vals(data_frame=input_df,
class_column='Response',
features=features)
new_feature_list = []
for feature in features:
for bin_val in range(1, n_bins + 1):
new_feature_list.append(feature + str(bin_val))
discr_arr = obj1.discretize_features(data_frame=no_zero_vals,
features=features,
n_bins=n_bins)
discr_df = pd.DataFrame(discr_arr, columns=new_feature_list)
discr_df['Response'] = dframe['Response']
splits = obj1.split_train_test(
x_data=discr_df[new_feature_list].to_numpy(),
y_labels=discr_df['Response'].to_numpy(),
features=new_feature_list,
test_size=0.25)
return splits
def test_load_data():
"""Test load_data()."""
obj = DataWrangling(data_path='src/data/data.csv')
obj.load_data()
assert all(obj.data.columns == pd.read_csv('src/data/data.csv').columns)