def test_cross_validate_n_predictors():
data_path = "./formulation/data/"
data_fname = 'FDA_APPROVED.csv'
# Read csv file
df = pd.read_csv(data_path + data_fname)
print(df.tail(3))
# Dropnan
columns = ['CLogP', 'HBA', 'HBD', 'PSDA', 'Formulation']
df = data_dropna(df, needed_cols=columns, subset=columns)
print(df.tail(10))
# Print count of each category
print(df.groupby('Formulation').count())
# Prepare predictors and response variable
features = ['CLogP', 'HBA', 'HBD', 'PSDA']
X_df = df[features]
target = ['Formulation']
y_df = df[target]
max_depth = 2
n_estimators = 100
best_p = cross_validate_n_predictors(X_df, y_df, max_depth, n_estimators)
assert isinstance(best_p, int)
assert best_p <= X_df.shape[1]
def test_cross_validate_grid_search():
data_path = "./formulation/data/"
data_fname = 'FDA_APPROVED.csv'
# Read csv file
df = pd.read_csv(data_path + data_fname)
print(df.tail(3))
# Dropnan
columns = ['CLogP', 'HBA', 'HBD', 'PSDA', 'Formulation']
df = data_dropna(df, needed_cols=columns, subset=columns)
print(df.tail(10))
# Print count of each category
print(df.groupby('Formulation').count())
# Prepare predictors and response variable
features = ['CLogP', 'HBA', 'HBD', 'PSDA']
X_df = df[features]
target = ['Formulation']
y_df = df[target]
max_depth = range(1, 5)
ntrees = range(1, 200, 50)
values = [max_depth, ntrees]
results = cross_validate_grid_search(values, X_df, y_df)
assert len(results) == 4
assert len(results[0]) == 2
assert len(results[1]) == 2
assert len(results[2]) == 2
assert len(results[3]) == 2
m = len(results)
n = len(results[0])
for i in range(m):
for j in range(n):
assert isinstance(results[i][j], int)
def test_predict():
data_path = "./formulation/data/"
data_fname = 'FDA_APPROVED.csv'
# Read csv file
df = pd.read_csv(data_path + data_fname)
print(df.tail(3))
# Dropnan
columns = ['CLogP', 'HBA', 'HBD', 'PSDA', 'Formulation']
df = data_dropna(df, needed_cols=columns, subset=columns)
print(df.tail(10))
# Print count of each category
print(df.groupby('Formulation').count())
# Prepare predictors and response variable
features = ['CLogP', 'HBA', 'HBD', 'PSDA']
X_df = df[features]
target = ['Formulation']
y_df = df[target]
X_trn, X_tst, y_trn, y_tst = train_test_split(X_df.values,
y_df.values,
test_size=0.20,
random_state=42)
clf = RandomForestClassifier(n_estimators=100, max_depth=2, random_state=0)
clf.fit(X_trn, y_trn.flatten())
y_pred = predict(clf, X_tst)
assert isinstance(y_pred, np.ndarray)
assert y_pred.shape[0] == y_tst.shape[0]
X_tst_df = pd.DataFrame(X_tst)
y_pred2 = predict(clf, X_tst_df)
assert isinstance(y_pred2, np.ndarray)
assert y_pred2.shape[0] == y_tst.shape[0]
def test_data_dropna():
"""
Test case for `data_dropna`
"""
# test handling illegal inputs
try:
# `data` should be a pd.DataFrame
data_dropna([1, 2, None])
raise Exception("Illegal input test failed!")
except AssertionError:
pass
try:
# `subset` contains elements that are not in `needed_cols`
data_dropna(DATA, INPUTS, NEEDED)
raise Exception("Illegal input test failed!")
except AssertionError:
pass
all_dropped = data_dropna(DATA)
inputs_dropped = data_dropna(DATA, NEEDED, INPUTS)
needed_dropped = data_dropna(DATA, NEEDED, NEEDED)
assert isinstance(all_dropped, pd.DataFrame),\
"Type error. pd.DataFrame expected"
assert isinstance(inputs_dropped, pd.DataFrame),\
"Type error. pd.DataFrame expected"
assert isinstance(needed_dropped, pd.DataFrame),\
"Type error. pd.DataFrame expected"
assert True not in all_dropped.isna().to_numpy(),\
"Error. Missing value not dropped"
assert True not in inputs_dropped[INPUTS].isna().to_numpy(),\
"Error. Missing value not dropped"
assert True not in needed_dropped.isna().to_numpy(),\
"Error. Missing value not dropped"
return best_p
if __name__ == '__main__':
data_path = "./formulation/data/"
data_fname = 'FDA_APPROVED.csv'
# Read csv file
df = pd.read_csv(data_path + data_fname)
print(df.tail(3))
# Dropnan
columns = ['CLogP', 'HBA', 'HBD', 'PSDA', 'Formulation']
df = data_dropna(df, needed_cols=columns, subset=columns)
print(df.tail(10))
# Print count of each category
print(df.groupby('Formulation').count())
# Prepare predictors and response variable
features = ['CLogP', 'HBA', 'HBD', 'PSDA']
X = df[features]
target = ['Formulation']
y = df[target]
max_depth = range(1, 5)
ntrees = range(1, 200, 50)