如何在.format中使用np.select？

Question

我有df如下所示，我正在尝试逐行生成一个状态列，其中包含特定的日期。

但是我得到了有关DF对象的信息，而不是行的值。

代码：

​

​

import pandas as pd
import numpy as np

data = {'local':  ['client', 'hub'],
    'delivery_date':  [pd.to_datetime('2022-04-24'), 'null'],
    'estimated_date':  [pd.to_datetime('2022-04-24'), pd.to_datetime('2022-04-26')],
    'max_date': ['delivery_date', 'estimated_date']
    }

 df = pd.DataFrame(data)
 cond = [(df["max_date"] == "delivery_date")  & (df["local"] == "client"),
         (df["max_date"] == "estimated_date") & (df["local"] == "hub")]

 choices = ["It was delivered to the customer on the date {}".format(df["delivery_date"]),
            "delivery forecast for {}".format(df["estimated_date"])]

 df["status"] = np.select(cond, choices, default = np.nan)

预期结果：

​

Answer 1

您可以使用字符串连接：

choices = ["It was delivered to the customer on the date " + df["delivery_date"].astype(str),
            "delivery forecast for " + df["estimated_date"].astype(str)]
df["status"] = np.select(cond, choices, default = np.nan)

As @Parfait注意到，如果您有pandas>=1.0.0，请为引入的StringDtype使用astype("string")。所以

choices = ["It was delivered to the customer on the date " + df["delivery_date"].astype('string'),
            "delivery forecast for " + df["estimated_date"].astype('string')]

输出：

    local        delivery_date estimated_date        max_date                                             status  
0  client  2022-04-24 00:00:00     2022-04-24   delivery_date  It was delivered to the customer on the date 2...   
1     hub                 null     2022-04-26  estimated_date                   delivery forecast for 2022-04-26  

Answer 2

由于Series对象包含许多值，而且str.format需要一个单数值，所以考虑string与的串联。

...

# ADD HELPER COLUMNS
df["delivery_note"] = "It was delivered to the customer on the date "
df["forecast_note"] = "delivery forecast for " 

choices = [
    df["delivery_note"].str.cat(df["delivery_date"]),
    df["forecast_note"].str.cat(df["estimated_date"])
]

df["status"] = np.select(cond, choices, default = np.nan)

# REMOVE HELPER COLUMNS
df.drop(["delivery_note", "forecast_note"], axis="columns", inplace=True)