二、字元串與整數拼接速度對比（Python中的AI對比實驗）

• 2020 年 4 月 22 日
• AI
• Python, PyTorch

完整程式碼如下：

# -*- coding: UTF-8 -*-

# Author: Perry
# @Create Time: 2020-04-06 14:55

import timeit
import numpy as np


def ints():
    _ = "str" + str(12)


def ints_f():
    _ = "str%d" % 12


def floats():
    _ = "str" + str(12.33)


def floats_f():
    _ = "str%f" % 12.33


def strs():
    _ = "str" + "12.33"


def strs_f():
    _ = "str%s" % "12.33"


if __name__ == '__main__':
    repeat = 10
    number = 100000

    int_time = timeit.repeat('ints()', 'from __main__ import ints', repeat=repeat, number=number)
    int_time = np.mean(int_time)
    print("int_time: ", int_time)

    int_f_time = timeit.repeat('ints_f()', 'from __main__ import ints_f', repeat=repeat, number=number)
    int_f_time = np.mean(int_f_time)
    print("int_f_time: ", int_f_time)

    float_time = timeit.repeat('floats()', 'from __main__ import floats', repeat=repeat, number=number)
    float_time = np.mean(float_time)
    print("float_time: ", float_time)

    float_f_time = timeit.repeat('floats_f()', 'from __main__ import floats_f', repeat=repeat, number=number)
    float_f_time = np.mean(float_f_time)
    print("float_f_time: ", float_f_time)

    str_time = timeit.repeat('strs()', 'from __main__ import strs', repeat=repeat, number=number)
    str_time = np.mean(str_time)
    print("str_time: ", str_time)

    str_f_time = timeit.repeat('strs_f()', 'from __main__ import strs_f', repeat=repeat, number=number)
    str_f_time = np.mean(str_f_time)
    print("str_f_time: ", str_f_time)

輸出結果：

int_time:  0.02541184720000001
int_f_time:  0.01889225879999996
float_time:  0.04497297539999997
float_f_time:  0.0266983679
str_time:  0.006410686699999979
str_f_time:  0.01752534760000002

結論：當有類型轉換的時候，使用 % 轉換更快；否則，若為str，則用+更快。