使用列组合查找行熊猫中的数据不匹配

Question

What's the best way to get all cell values based on a combination of column values?

Sample dataframe One:

  Stock                         Name  Price
0    AMD       Advanced Micro Devices    100
1     GE     General Electric Company    200
2    BAC  Bank of America Corporation    300
3   AAPL                   Apple Inc.    500
4   MSFT        Microsoft Corporation   1000
5  GOOGL                Alphabet Inc.   2000

Sample dataframe Two:

  Stock                         Name  Price
0    AMD       Advanced Micro Devices    100
1     GE     General Electric Company    200
2    BAC  Branch of America Corporation  300
3   AAPL                   Apple Inc.    500
4   MSFT        Microsoft Corporation   1000
5  GOOGL                Alphabet Inc.   2000

For example: I want to use (Stock and Name) as key columns and then compare the datasets. The goal is to print the mismatch entries between the two datasets with the Stock+Name columns used as a combination key.

I'm using Pandas/Python3.7

Sample Output:

BAC Bank of America Corporation 300 --- BAC Branch of America Corporation 300

Answer 1

Perhaps, a FULL INNER JOIN using merge + query ?

df1.merge(df2, on='Stock').query('Name_x != Name_y')

  Stock                       Name_x  Price_x                         Name_y  Price_y
2   BAC  Bank of America Corporation      300  Branch of America Corporation      300

---

Or, a slightly different solution with map , you can use to get the stock symbols:

m = df1.Stock.map(df2.set_index('Stock').Name).ne(df1.Name)
symbols = df1.loc[m, 'Stock']

print(symbols)
2    BAC
Name: Stock, dtype: object

And then access each DataFrame row by stock symbol:

df1[df1.Stock.isin(symbols)]
  Stock                         Name  Price
2   BAC  Bank of America Corporation    300

df2[df2.Stock.isin(symbols)]
  Stock                           Name  Price
2   BAC  Branch of America Corporation    300

Answer 2

If they are in two dataframes, merging them without condition is pretty straightforward with .concat . Once they are joined, here's one way to get the mismatch:

import pandas as pd

df1 = pd.DataFrame({
    "Ticker_y": list("qwerty"),
    "Name_y": list("asdfgh"),
    "Ticker_x": list("qw3r7y"),
    "Name_x": list("as6f8h")
})

mismatch = df1[(df1["Ticker_y"] != df1["Ticker_x"]) & (df1["Name_y"] != df1["Name_x"])]

The last line just says "the df only where these conditions are met."

Answer 3

We can use isin using the sequence of values to test as it ensures each element in the DataFrame is contained in values

First DataFrame

>>> df1
   Stock                         Name  Price
0    AMD       Advanced Micro Devices    100
1     GE     General Electric Company    200
2    BAC  Bank of America Corporation    300
3   APPL                   Apple Inc.    500
4   MSFT        Microsoft Corporation   1000
5  GOOGL                Alphabet Inc.   2000

Second DataFrame

>>> df2
   Stock                           Name  Price
0    AMD         Advanced Micro Devices    100
1     GE       General Electric Company    200
2    BAC  Branch of America Corporation    300
3   APPL                     Apple Inc.    500
4   MSFT          Microsoft Corporation   1000
5  GOOGL                  Alphabet Inc.   2000

Here you can go..

>>> df2[~df2.Name.isin(df1.Name.values)]
  Stock                           Name  Price
2   BAC  Branch of America Corporation    300

OR

>>> df1[~df1.Name.isin(df2.Name.values)]
  Stock                         Name  Price
2   BAC  Bank of America Corporation    300