从其他 Dataframe 更新多个 Dataframe 列

Question

I have two dataframes:

df1

         Date         ID Company     Symbol  Shares  Value
0  2014-09-30   32511907     foo        NaN      10    101
1  2014-09-30   32511957     bar  No Symbol      10    101
2  2014-12-31   32511907    test  No Symbol      18    152
3  2013-06-30   32511107      AA        APC      43    373
4  2014-06-30  166764100      CC        CVX      29    381
5  2014-09-30  166764900       C  No Symbol      13    155
6  2014-09-30  166764950   C Inc  No Symbol      13    155
7  2015-09-30  475643276    Test        NaN      13    155

And df2 :

   Company         ID     Symbol
0   AA Inc   32511907        AAA
1   AA Inc   32511957  No Symbol
2   BB Inc   32511107        APC
3  CC Corp  166764100        CVX
4   CC Inc  166764900  No Symbol
5   CC Inc  166764950  No Symbol

By using ID as the key, I need to replace Symbol and Company in df1 , with those in df2 , so that the output looks like:

         Date         ID Company     Symbol  Shares  Value
0  2014-09-30   32511907  AA Inc        AAA      10    101
1  2014-09-30   32511957  AA Inc  No Symbol      10    101
2  2014-12-31   32511907  AA Inc        AAA      18    152
3  2013-06-30   32511107  BB Inc        APC      43    373
4  2014-06-30  166764100 CC Corp        CVX      29    381
5  2014-09-30  166764900  CC Inc  No Symbol      13    155
6  2014-09-30  166764950  CC Inc  No Symbol      13    155
7  2015-09-30  475643276    Test        NaN      13    155

I have tried:

df = df1.set_index('ID')
df.update(df2.set_index('ID'))
df = df1.loc[df1.Symbol == '', ['Company', 'Symbol']] = df.reset_index()

But when performed on the full dataset, I get ValueError: cannot reindex from a duplicate axis because line df.update(df2.set_index('ID')) makes ID the index, which results in a few duplicates.

So are there any alternatives to achieving the desired output?

Answer 1

A pd.merge can solve this. Assume you only want IDs in df1

result = pd.merge(left=df1[['Date', 'ID', 'Shares', 'Values']], 
                  right=df2, on=['ID'], how='left')

If you need to merge on different labels, you can use left_on and right_on.

Change the how argument to specify what IDs to keep

• keep IDs only occurred in df1 ('left')
• keep IDs only occurred in df2 ('right')
• keep all IDs occurred in both df1 and df2 ('outer')
• keep only IDs occurred in both df1 and df1 ('inner')