如何加速基于列表统计的共识判断Python代码？

我正在优化一个共识判断函数，它接收字符串列表（比如['bull','bull','bear']）作为输入，需要得出“多数共识”，输出字符串'bull'。想请教如何提升它的运行速度。

原代码 #

consensus_type = "majority"
activeIndicators = ["bull","bull","bear"]

def consensus(
    activeIndicators
    ):

    counterBull = activeIndicators.count("bull")
    counterBear = activeIndicators.count("bear")
    counterNeutral = activeIndicators.count("neutral")

    lists = [counterBull,counterBear,counterNeutral]
    max_value_from_list = max(lists)
    count_of_max_value_in_list = lists.count(max_value_from_list)

    if consensus_type == "majority":
        if count_of_max_value_in_list == 1:
            d = {'bull':counterBull,'bear':counterBear,'neutral':counterNeutral}
            consensus = max(d, key=d.get)
        else:
            consensus = "neutral"
    
    elif consensus_type == "unanimity":
        if max_value_from_list == len(activeIndicators):
            d = {'bull':counterBull,'bear':counterBear,'neutral':counterNeutral}
            consensus = max(d, key=d.get)
        else:
            consensus = "neutral"

    return consensus

已完成的优化（速度提升至原版本两倍） #

以下是简化后的优化代码，目前运行速度是原版本的两倍，希望得到更多优化建议：

def consensus(
    activeIndicators
    ):

    counterBull = activeIndicators.count("bull")
    counterBear = activeIndicators.count("bear")
    counterNeutral = len(activeIndicators) - counterBull - counterBear

    consensus = "neutral"

    if counterBull >= counterBear and counterBull >= counterNeutral:
        if consensus_type == 'unanimity' and counterBull == len(activeIndicators):
            consensus = "bull"
        elif consensus_type == 'majority' and (counterBull != counterBear and counterBull != counterNeutral):
            consensus = "bull"

    elif counterBear >= counterBull and counterBear >= counterNeutral:
        if consensus_type == 'unanimity' and counterBear == len(activeIndicators):
            consensus = "bear"
        elif consensus_type == 'majority' and (counterBear != counterBull and counterBear != counterNeutral):
            consensus = "bear"

    return consensus

进一步优化建议 #

1. 单次遍历完成计数，减少列表遍历次数 #

原代码和当前优化代码都调用了两次count()方法，每次count()都会完整遍历列表一次。改成单次遍历统计所有类型数量，能直接减少一半的遍历开销，列表越长收益越明显：

def consensus(activeIndicators, consensus_type="majority"):
    counterBull = 0
    counterBear = 0
    for item in activeIndicators:
        if item == "bull":
            counterBull += 1
        elif item == "bear":
            counterBear += 1
    counterNeutral = len(activeIndicators) - counterBull - counterBear

    consensus = "neutral"
    if counterBull >= counterBear and counterBull >= counterNeutral:
        if consensus_type == 'unanimity' and counterBull == len(activeIndicators):
            consensus = "bull"
        elif consensus_type == 'majority' and (counterBull != counterBear and counterBull != counterNeutral):
            consensus = "bull"
    elif counterBear >= counterBull and counterBear >= counterNeutral:
        if consensus_type == 'unanimity' and counterBear == len(activeIndicators):
            consensus = "bear"
        elif consensus_type == 'majority' and (counterBear != counterBull and counterBear != counterNeutral):
            consensus = "bear"
    return consensus

2. 将consensus_type作为函数参数传入 #

当前代码中consensus_type是全局变量，既降低了函数的封装性，全局变量的访问速度也略慢于局部变量。把它作为参数传入，不仅让函数更易复用，还能小幅提升性能。

3. 提前按consensus_type分支，减少嵌套层级 #

调整判断逻辑顺序，先根据consensus_type分支，再处理计数后的判断，既能让代码逻辑更清晰，也能减少运行时的嵌套条件判断开销：

def consensus(activeIndicators, consensus_type="majority"):
    counterBull = 0
    counterBear = 0
    for item in activeIndicators:
        if item == "bull":
            counterBull += 1
        elif item == "bear":
            counterBear += 1
    counterNeutral = len(activeIndicators) - counterBull - counterBear

    if consensus_type == 'unanimity':
        if counterBull == len(activeIndicators):
            return "bull"
        elif counterBear == len(activeIndicators):
            return "bear"
        return "neutral"
    elif consensus_type == 'majority':
        if counterBull > counterBear and counterBull > counterNeutral:
            return "bull"
        elif counterBear > counterBull and counterBear > counterNeutral:
            return "bear"
        return "neutral"

4. 处理长列表时使用collections.Counter #

如果需要处理的列表规模很大，使用标准库的collections.Counter会更高效——它的底层实现是C语言，统计速度比纯Python遍历更快：

from collections import Counter

def consensus(activeIndicators, consensus_type="majority"):
    cnt = Counter(activeIndicators)
    counterBull = cnt.get("bull", 0)
    counterBear = cnt.get("bear", 0)
    counterNeutral = cnt.get("neutral", 0)

    if consensus_type == 'unanimity':
        if counterBull == len(activeIndicators):
            return "bull"
        elif counterBear == len(activeIndicators):
            return "bear"
        return "neutral"
    elif consensus_type == 'majority':
        if counterBull > counterBear and counterBull > counterNeutral:
            return "bull"
        elif counterBear > counterBull and counterBear > counterNeutral:
            return "bear"
        return "neutral"

注意：如果列表很短，引入Counter的初始化开销可能超过收益，因此更适合大规模数据场景。

内容的提问来源于stack exchange，提问作者zolp