python - 如何使此功能更有效？（板球统计）

Question

我正在编写一个函数batting_stats来显示击球手职业生涯中每一局的击球统计数据。该函数的输入是一个整数列表。顶级列表包含一局列表，而内部列表包含得分、面对的球和一个布尔值（1 = 出局或 0 = 未出局），指示击球手是否在局中被解雇。该函数是返回一个整数列表，表示击球手的平均数、击球率和每一局的转换率。

平均= 跑数 // 解雇次数（如果玩家没有被解雇，则平均值为总得分）

罢工率= 100*runs // 面对的球

转换率= 100 * 得分的世纪数 / 50+ 得分的数量。（如果玩家没有得分大于 50，则转换率为零）

Input Format:
[[r1,b1,d1],[r2,b2,d2],...]
where r=runs, b=balls, d=dissmissal

Output Format:
[[avg1,sr1,cr1],[avg2,sr2,cr2],...]
where avg=average, sr=strike rate, cr=conversion rate

例如：

>>> batting_stats([[12,24,0],[18,36,1]])
[[12,50,0],[30,50,0]]

我的代码给了我预期的结果，但显然实现不是最优的。对于非常大的输入，我遇到了超时错误。我该如何优化它？

def batting_stats(lst):
    """Compute the average, strike rate, and conversion rate of a batsman after each innings."""
    innings = len(lst)  # number of innings
    last = 1 + innings
    r_lst = [r[0] for r in lst]  # list of runs per innings
    b_lst = [b[1] for b in lst]  # list of balls faced per innings
    d_lst = [d[2] for d in lst]  # list of dismissals per innings
    c_lst = [1 if r >= 100 else 0 for r in r_lst]  # list of 100+ scores
    f_lst = [1 if r >= 50 else 0 for r in r_lst]  # list of 50+ scores

    # Keep track of sums after each innings
    rt = [sum(r_lst[:n]) for n in range(1, last)]  # runs scored
    bt = [sum(b_lst[:n]) for n in range(1, last)]  # balls faced
    dt = [sum(d_lst[:n]) for n in range(1, last)]  # dismissals
    ct = [sum(c_lst[:n]) for n in range(1, last)]  # 100+ scores
    ft = [sum(f_lst[:n]) for n in range(1, last)]  # 50+ scores

    avg_ = [rt[i] if dt[i] == 0 else rt[i] // dt[i] for i in range(innings)]  # averages after each innings
    sr_ = [100 * rt[i] // bt[i] for i in range(innings)]  # strike rates after each innings
    cr_ = [0 if ft[i] == 0 else 100 * ct[i] // ft[i] for i in range(innings)]  # conversion rates after each innings

    return [[avg_[i], sr_[i], cr_[i]] for i in range(innings)]

标签： pythonlistfunctionoptimization

Answer 1

我结合了 Daniel Mesejo 的建议（使用itertools.accumulate）和unzip模块more_itertools来修改您的功能（请原谅重新格式化/重命名 - 这是为了我自己的可读性）。我还对zip.

def batting_stats2(lst):
    """Compute the average, strike rate, and conversion rate of a batsman after each innings."""
    innings = len(lst)  # number of innings
    # Not needed
    #last = 1 + innings

    # Unzip reshapes the various stats into their own lists
    r_lst, b_lst, d_lst = map(list, unzip(lst))

    # Similarly, for the 100+ and 50+ scores
    # Note: int(True) = 1, int(False) = 0
    c_lst, f_lst = map(list, unzip((int(r >= 100), int(r >= 50)) for r in r_lst))

    # Accumulate the sums
    rt = list(accumulate(r_lst)) # list of runs per innings
    bt = list(accumulate(b_lst)) # list of balls faced per innings
    dt = list(accumulate(d_lst)) # list of dismissals per innings
    ct = list(accumulate(c_lst)) # list of 100+ scores
    ft = list(accumulate(f_lst)) # list of 50+ scores

    # averages after each innings
    avg_ = [run if dismiss == 0 else run // dismiss for run, dismiss in zip(rt, dt)]

    # strike rates after each innings
    sr_ = [100 * run // ball for run, ball in zip(rt, bt)]

    # conversion rates after each innings
    cr_ = [fifty if fifty == 0 else 100 * hundo // fifty for fifty, hundo in zip(ft, ct)]

    # The "list(x)" is because your output is nested lists. Without, it would be
    # list of tuples.
    return list(list(x) for x in zip(avg_, sr_, cr_))

然后我用你发布的例子进行了测试，得到了相同的输出，所以它似乎计算正确。

之后，我输入了 1000000 个 3 成员列表（称为stats）来计时：

>>> %%timeit
>>> batting_stats2(stats)

2.43 s ± 35.3 ms per loop (mean ± std. dev. of 7 runs, 1 loop each)

我也试过给原版计时，但等了10分钟就放弃了：）