如何从迭代两个数据帧中提取特定值并动态地提取 Append？

Question

I am trying to perform a multi-step way of using values in an existing dataframe to create subset dataframes (or series) and compare those values against the existing dataframe to ultimately create one new column in the original dataframe.

The step by step process is:

• Create a dataframe that isolates the minimum date within each month (df2) based on dates in the original dataframe, df1, and this is successfully being created.
• The second step is to match where the actual_date (df1) matches to the the min_date (df2) and then take the extract_value (df1) and store it in df1 for all instances where the actual_date matches the min_date .

Some attempts I've taken but am receiving errors are: Comparing date values between dfs:

df1.loc[df1['actual_date']==df2[df2['min_date']]
#Produces unexpected EOF while parsing

df['actual_date']==df2['min_date']
#Produces ValueError: Can only compare identically-labeled Series objects

Iterating through conditions:

for each in range(len(df1):
    if df1[df1['actual_date']]==df2[df2['min_date']]:
        df1['exctract_value_new']=df2['extract_value']

#Produces: KeyError: "None of [DatetimeIndex....are in the [columns]"

I've tried searching both value and key errors and am having trouble understanding the threads regarding indexing. Specifically I am unsure how df1 and/or df2 need to be reformatted to compare date values in this manner and then extract a separate column based on when criteria matches in both dataframes.

Here is the sample data working with:

df1 (base)

actual_date	extract_value
2021-01-22	22
2021-01-23	24
2021-01-24	15
2021-02-22	16
2021-02-05	34
2021-02-04	18

df2

month	min_date
2021-01-01	2021-01-22
2021-02-01	2021-02-04
2021-03-01	2021-03-01

End Goal for df1

actual_date	min_date	extract_value_new	extract_value_original
2021-01-22	2021-01-22	22	22
2021-01-23	2021-01-22	22	24
2021-02-04	2021-02-04	18	18
2021-02-05	2021-02-04	18	34

Appreciate any help!

Answer 1

You might want to check out pd.merge_asof ...solution here gives me the desired table!

merge_asof essentially lets us join on the "closest" values in two columns.

import pandas as pd
import requests
from datetime import datetime, timedelta


df1 = pd.read_csv('df1.csv')
df2 = pd.read_csv('df2.csv')

datetime_format = '%Y-%m-%d'  # format of date (needs to be a datetime to merge in pandas)
tolerance = timedelta(days=1)  # the tolerance to join dates on

# convert our dates
df1['actual_date'] = pd.to_datetime(df1['actual_date'], format=datetime_format)
df2['min_date'] = pd.to_datetime(df2['min_date'], format=datetime_format)

# sort...
df1.sort_values(by='actual_date', inplace=True)

out = pd.merge_asof(df1, df2, left_on='actual_date', right_on='min_date', tolerance=tolerance).dropna()

Gives me:

  actual_date  extract_value       month   min_date
0  2021-01-22             22  2021-01-01 2021-01-22
1  2021-01-23             24  2021-01-01 2021-01-22
3  2021-02-04             18  2021-02-01 2021-02-04
4  2021-02-05             34  2021-02-01 2021-02-04