I have the following dataframes in Pandas:
df1:
index column
1 A1
2 A2
df2:
index column
2 A2_new
3 A3
I want to get the result:
index column
1 A1
2 A2_new
3 A3
How do I can achieve this?
df1.update(df2) is not helpful, because I want to see row with index 3 in the result.
Example
df1 = pd.DataFrame(['A1', 'A2'], columns=['column'], index=[1, 2])
df2 = pd.DataFrame(['A2_new', 'A3'], columns=['column'], index=[2, 3])
df1
column
1 A1
2 A2
df2
column
2 A2_new
3 A3
Code
df2.combine_first(df1)
output
column
1 A1
2 A2_new
3 A3
@Ars ML You can concatenate the two DataFrames vertically and remove duplicates from 'index' column, keeping only the last occurrence of each index value
df1 = pd.DataFrame({'index': [1, 2], 'column': ['A1', 'A2']})
df2 = pd.DataFrame({'index': [2, 3], 'column': ['A2_new', 'A3']})
merged_df = pd.concat([df1, df2]).drop_duplicates(subset=['index'], keep='last')
merged_df.set_index('index', inplace=True)
outputs as per your desired outcome.
1 A1
2 A2_new
3 A3
You can also use merge, it is more involved but produces your desired outcome.
merge_chain = pd.merge(df1, df2, on='index', how='outer') \
.assign(column=lambda x: x['column_y'].fillna(x['column_x'])) \
.drop(['column_x', 'column_y'], axis=1) \
.set_index('index')
Another possible solution:
out = pd.concat([df1, df2])
out[~out.index.duplicated(keep='last')]
Output:
column
1 A1
2 A2_new
3 A3
df1 = df2.reindex(df1.index.union(df2.index)).fillna(df1)work for your use case?