在熊猫组上迭代非常慢，并且应用功能无法正常工作

Question

我正在使用pandas为我们的客户生成自定义报告。生成报告的流程如下：

1. 从API获取数据
2. 通过pandas函数进行解析，执行一些业务逻辑，并通过email发送生成的报告。

业务逻辑： 我们必须在数据中找到所有人的first-in和last-out时间。

代码：

df = pd.DataFrame(report_data)
df["Code"] = df["[People]Employee Code"]
df["Checkin At Date"] = pd.to_datetime(df["Checkin At Date"])
df["Checkin At Time"] = pd.to_datetime(df["Checkin At Time"])
grouped_data = df.groupby(["Mobile Number", "Checkin At Date"])
final_df = pd.DataFrame(columns=df.columns)
for key, temp_df in grouped_data:
    try:
        num_of_rows = len(temp_df.index)
        temp_df.sort_values(
            by=["Checkin At Time"], inplace=True, ascending=True
        )
        first_checkin_time = temp_df.loc[
            temp_df.index[0], "Checkin At Time"
        ]
        last_checkin_time = temp_df.loc[
            temp_df.index[-1], "Checkin At Time"
        ]
        temp_df['FIRST_BODY_TEMP'] = temp_df.loc[
            temp_df.index[0], "Body Temp"
        ]
        temp_df['LAST_BODY_TEMP'] = temp_df.loc[
            temp_df.index[-1], "Body Temp"
        ]
        if num_of_rows <= 1:
            temp_df.loc['TOTAL_TIME'] = 0
            temp_df.loc['LAST_OUT_TIME'] = None
        else:
            temp_df['TOTAL_TIME'] = last_checkin_time - first_checkin_time
            temp_df['LAST_OUT_TIME'] = temp_df.loc[
                    temp_df.index[-1], "Checkin At Time"]

            temp_df["TOTAL_TIME"] = pd.to_timedelta(
                temp_df["TOTAL_TIME"], unit="s"
            )
            temp_df["TOTAL_TIME"] = temp_df["TOTAL_TIME"].apply(
                lambda x: strfdelta(x, "{hours} Hrs {minutes} Min")
            )
            temp_df["LAST_OUT_TIME"] = temp_df["LAST_OUT_TIME"].dt.time
        # Fetching the first row of the grouped data and appending to df
        final_df = final_df.append(temp_df.iloc[0], ignore_index=True)
    except Exception as e:
        logger.exception(e)
        continue

对于记录容量800 items，这几乎要用18-20 seconds。

我尝试过的事情：

def calculate_attendance(temp_df):
    num_of_rows = len(temp_df.index)
    temp_df.sort_values(by=['Checkin At Time'], inplace=True,
                    ascending=True)
    first_checkin_time = temp_df.loc[temp_df.index[0], 'Checkin At Time'
        ]
    last_checkin_time = temp_df.loc[temp_df.index[-1], 'Checkin At Time'
                                ]
    temp_df['FIRST_BODY_TEMP'] = temp_df.loc[temp_df.index[0],
        'Body Temp']
    temp_df['LAST_BODY_TEMP'] = temp_df.loc[temp_df.index[-1],
        'Body Temp']
    if num_of_rows <= 1:
       temp_df.loc['TOTAL_TIME'] = 0
       temp_df.loc['LAST_OUT_TIME'] = None
    else:
       temp_df['TOTAL_TIME'] = last_checkin_time - first_checkin_time
       temp_df['LAST_OUT_TIME'] = temp_df.loc[temp_df.index[-1],
            'Checkin At Time']

       temp_df['TOTAL_TIME'] = pd.to_timedelta(temp_df['TOTAL_TIME'],
            unit='s')
       temp_df['TOTAL_TIME'] = temp_df['TOTAL_TIME'].apply(lambda x: \
            strfdelta(x, '{hours} Hrs {minutes} Min'))
       temp_df['LAST_OUT_TIME'] = temp_df['LAST_OUT_TIME'].dt.time
       return temp_df.iloc[0]

final_df = pd.DataFrame(columns=df.columns)
final_df = final_df.groupby(["Mobile Number", "Checkin At Date"]).apply(calculate_attendance)

此步骤中的问题是：

同一组重复出现，final_df位于同一行。