def main():
y = np.random.randint(2, size=(5000, 1))
x = np.random.randint(10, size=(5000, 1))
data = pd.DataFrame(np.concatenate([y, x], axis=1), columns=['y', 'x'])
a = time.time()
result_1 = target_mean_v2(data, 'y', 'x')
print('最开始的第二种方法 ', time.time() - a)
a = time.time()
result_2 = target_mean_v3(data, 'y', 'x')
print('通过transform写的方法 ', time.time() - a)
a = time.time()
result_3 = tm.target_mean_v3(data, 'y', 'x')
print('王然老师的方法 ', time.time() - a)
# 可以改的思路：unordered_map, 数据类型:int, float; 该循环 for row from 0 <= row < nrow by 1:
a = time.time()
result_4_type_change, value_dict, count_dict = tm.target_mean_v4(
data, 'y', 'x')
# print(result_4, value_dict, count_dict)
print('改写数据类型的方法 ', time.time() - a)
a = time.time()
result_4_unordered_map, value_dict, count_dict = tm.target_mean_v4_unordered_map(
data, 'y', 'x')
print('修改成unordered_map ', time.time() - a)
print(np.linalg.norm(result_2 - result_1))
print(np.linalg.norm(result_3 - result_1))
print(np.linalg.norm(result_4_type_change - result_1))
print(np.linalg.norm(result_4_unordered_map - result_1))
def main():
y = np.random.randint(2, size=(5000, 1))
x = np.random.randint(10, size=(5000, 1))
data = pd.DataFrame(np.concatenate([y, x], axis=1), columns=['y', 'x'])
start = time.time()
result = tm.target_mean_v3(data, 'y', 'x')
end = time.time()
print(end - start)
def main():
size = 100000
print(f'{size} test data start in {time()}, please wait.')
y = np.random.randint(2, size=(size, 1))
x = np.random.randint(10, size=(size, 1))
data = pd.DataFrame(np.concatenate([y, x], axis=1), columns=['y', 'x'])
start_2 = time()
target_mean_v2(data, 'y', 'x')
end_2 = time()
print(f'v2 is the python version, use time: {end_2 - start_2}')
start_3 = time()
tm.target_mean_v3(data, 'y', 'x')
end_3 = time()
print(f'v3 is the version showed by Mr.Wang, use time: {end_3 - start_3}')
start_4 = time()
tm.target_mean_v4(data, 'y', 'x')
end_4 = time()
print(f'v4 is my job, use time: {end_4 - start_4}')
def main():
y = np.random.randint(2, size=(5000, 1))
x = np.random.randint(10, size=(5000, 1))
data = pd.DataFrame(np.concatenate([y, x], axis=1), columns=['y', 'x'])
start = time.time()
result_1 = target_mean_v1(data, 'y', 'x')
print("1", time.time() - start)
start = time.time()
result_2 = tm.target_mean_v2(data, 'y', 'x')
print("2", time.time() - start)
start = time.time()
result_3 = tm.target_mean_v3(data, 'y', 'x')
print("3", time.time() - start)
print(np.linalg.norm(result_1 - result_2))
print(np.linalg.norm(result_2 - result_3))
def main():
y = np.random.randint(2, size=(5000, 1))
x = np.random.randint(10, size=(5000, 1))
data = pd.DataFrame(np.concatenate([y, x], axis=1), columns=['y', 'x'])
start = time()
result_1 = target_mean_v1(data, 'y', 'x')
end = time()
print("v1: {}".format(end-start))
result_2 = target_mean_v2(data, 'y', 'x')
end2 = time()
print("v2: {}".format(end2-end))
result_3 = tm.target_mean_v3(data, 'y', 'x')
end3 = time()
print("v3: {}".format(end3 - end2))
result_5 = tm.target_mean_v5(data, 'y', 'x')
end4 = time()
print("v5: {}".format(end4 - end3))
result_4 = tm.target_mean_v4(data, 'y', 'x')
end5 = time()
print("v4: {}".format(end5 - end4))
def call_tm_target_mean_v3(data, y_name, x_name):
start_time = time.time()
result = tm.target_mean_v3(data, y_name, x_name)
end_time = time.time()
print('cython实现tm.target_mean_v3执行时间： ', end_time - start_time)