使用ThreadPoolExecutor获取数据后，如何正确将列表作为行添加至DataFrame？

原代码 #

#!/usr/bin/env python
import sys
import yfinance as yf
from concurrent.futures import ThreadPoolExecutor
import pandas as pd

def get_stats(ticker):
    info = yf.Tickers(ticker).tickers[ticker].info
    data = [ticker, info['volume']/info['floatShares'], info['floatShares']]
    return data

ticker_list = sys.argv[1:]
ticker_list = ['AAPL', 'AMZN', 'GOOG']
ticker_list = [item.upper() for item in ticker_list]

table_title = ['Symbol', 'FTR', 'Float']
df = pd.DataFrame(columns = table_title)

with ThreadPoolExecutor() as executor:
    results = list(executor.map(get_stats, ticker_list))
    df.loc[len(df)] = results

print(df)

实际输出 #

Symbol                                       FTR                                       Float
0  [AAPL, 0.003960062654206358, 15179506298]  [AMZN, 0.006952834770567285, 9360261411]  [GOOG, 0.0029524428963116282, 10942799280]

期望输出 #

Symbol                   FTR        Float
0   AAPL  0.003960062654206358  15179506298
1   AMZN  0.006952834770567285   9360261411
2   GOOG  0.002952442896311628  10942799280

问题原因 #

df.loc[len(df)] = results 是将整个results列表（包含3个股票的子列表）作为一行数据赋值给DataFrame，导致每个列都被填充了一个完整的子列表，而非将每个子列表作为单独的一行。

修正方案 #

直接用results二维列表构造DataFrame，无需先创建空DataFrame再逐行添加：

#!/usr/bin/env python
import sys
import yfinance as yf
from concurrent.futures import ThreadPoolExecutor
import pandas as pd

def get_stats(ticker):
    info = yf.Tickers(ticker).tickers[ticker].info
    data = [ticker, info['volume']/info['floatShares'], info['floatShares']]
    return data

ticker_list = sys.argv[1:]
ticker_list = ['AAPL', 'AMZN', 'GOOG']
ticker_list = [item.upper() for item in ticker_list]

table_title = ['Symbol', 'FTR', 'Float']

with ThreadPoolExecutor() as executor:
    results = list(executor.map(get_stats, ticker_list))
    # 直接用results构造DataFrame，每个子列表对应一行
    df = pd.DataFrame(results, columns=table_title)

print(df)

补充说明 #

如果需要保留逐行添加的逻辑（比如需要处理中间数据），可以循环遍历results逐个添加：

df = pd.DataFrame(columns = table_title)
with ThreadPoolExecutor() as executor:
    results = list(executor.map(get_stats, ticker_list))
    for row in results:
        df.loc[len(df)] = row

内容的提问来源于stack exchange，提问作者Tex