def predict(data):
"""
Predicts the fitted model. Takes as an argument raw data, preprocesses it and does test-train splitting
Args:
data: directory with raw data without any preprocessing or test train splitting
Returns:
result_xgb(int64): predicted count of bikes
"""
hour_d = whole_preprocessing(data)
hour_d_train_x, hour_d_train_y, hour_d_test_x, hour_d_test_y = test_train_split(
hour_d)
try:
result_xgb = clf.predict(hour_d_test_x)
print("R-squared for Train: %.2f" %
clf.score(hour_d_train_x, hour_d_train_y))
print("R-squared for Test: %.2f" %
clf.score(hour_d_test_x, hour_d_test_y))
RMSE = np.sqrt(np.mean((hour_d_test_y * 2 - result_xgb * 2)**2))
MSE = RMSE**2
print("MSE ={}".format(MSE))
print("RMSE = {}".format(RMSE))
except:
try:
clf = load("model.pkl")
result_xgb = clf.predict(hour_d_test_x)
print("R-squared for Train: %.2f" %
clf.score(hour_d_train_x, hour_d_train_y))
print("R-squared for Test: %.2f" %
clf.score(hour_d_test_x, hour_d_test_y))
RMSE = np.sqrt(np.mean((hour_d_test_y * 2 - result_xgb * 2)**2))
MSE = RMSE**2
print("MSE ={}".format(MSE))
print("RMSE = {}".format(RMSE))
except:
try:
clf = train_and_persist(hour_d_train_x, hour_d_train_y,
hour_d_test_x, hour_d_test_y)
result_xgb = clf.predict(hour_d_test_x)
print("R-squared for Train: %.2f" %
clf.score(hour_d_train_x, hour_d_train_y))
print("R-squared for Test: %.2f" %
clf.score(hour_d_test_x, hour_d_test_y))
RMSE = np.sqrt(np.mean(
(hour_d_test_y * 2 - result_xgb * 2)**2))
MSE = RMSE**2
print("MSE ={}".format(MSE))
print("RMSE = {}".format(RMSE))
except:
pass
return "predictions-available"
def test_train_split(data=whole_preprocessing(loading_data())):
hour_d_train = data.iloc[0:15211]
hour_d_test = data.iloc[15212:17379]
hour_d_train = hour_d_train.drop(
columns=["dteday", "casual", "atemp", "registered"])
hour_d_test = hour_d_test.drop(
columns=["dteday", "casual", "registered", "atemp"])
# seperate the independent and target variable on testing data
hour_d_train_x = hour_d_train.drop(columns=["cnt"], axis=1)
hour_d_train_y = hour_d_train["cnt"]
# seperate the independent and target variable on test data
hour_d_test_x = hour_d_test.drop(columns=["cnt"], axis=1)
hour_d_test_y = hour_d_test["cnt"]
return hour_d_train_x, hour_d_train_y, hour_d_test_x, hour_d_test_y
def test_predict_data_cols_len():
output = len(whole_preprocessing(loading_data()).columns)
expected_output = 85
assert output == expected_output
def test_predict_data_rows_len():
output = len(whole_preprocessing(loading_data()))
expected_output = 17379
assert output == expected_output
def test_test_train_split(data=whole_preprocessing(loading_data())):
output = test_train_split(data)
assert type(output) is tuple
def test_test_train_split_train_not_cnt(data=whole_preprocessing(
loading_data())):
output = test_train_split(data)[2].columns
expected_output = "cnt"
assert expected_output not in output
def test_test_train_split_target_type(data=whole_preprocessing(
loading_data())):
cnt = test_train_split(data)[1]
output = type(cnt)
expected_output = pd.core.series.Series
assert output == expected_output
def test_test_train_split_column_length(data=whole_preprocessing(
loading_data())):
col_num = test_train_split(data)[0]
output = len(col_num.columns)
expected_output = 80
assert output == expected_output
def test_test_train_split_length(data=whole_preprocessing(loading_data())):
output = len(test_train_split(data)[0])
expected_output = 15211
assert output == expected_output
def test_test_train_split_length_tuple(data=whole_preprocessing(
loading_data())):
output = len(test_train_split(data))
expected_output = 4
assert output == expected_output
