I have a dataframe that looks like this:
fruit cost
apples 2
oranges 3
bananas 5
grapefruit 1
I want to pass a list that includes specified "fruit" column values and use that to duplicate those rows in the dataframe. For example, duplicated fruits = ['apples', 'oranges'].
These rows should then be copied back into the dataframe with an extra column that denotes that they are a copy (can be a binary 1/0).
Use Series.isin for get matching rows and for duplication use DataFrame.append to original data with DataFrame.assign for indicato column:
duplicated = ['apples', 'oranges']
df1 = df[df['fruit'].isin(duplicated)].assign(new=1)
df = df.assign(new=0).append(df1, ignore_index=True)
print (df)
fruit cost new
0 apples 2 0
1 oranges 3 0
2 bananas 5 0
3 grapefruit 1 0
4 apples 2 1
5 oranges 3 1
Another idea is use parameter keys in concat - it create new level filled by 0 and 1, so necessary DataFrame.reset_index by first level for convert this level for column:
df = (pd.concat([df, df1], keys=(0,1))
.rename_axis(('new', None))
.reset_index(level=0)
.reset_index(drop=True))
print (df)
new fruit cost
0 0 apples 2
1 0 oranges 3
2 0 bananas 5
3 0 grapefruit 1
4 1 apples 2
5 1 oranges 3
m =(df['fruit'] =='apples') & (df['new'] ==1) and then df.loc[m, 'fruit'] = "apple juice"You could use concat:
result = pd.concat([df, df[df.fruit.isin(fruits)].assign(new=1)], sort=False).fillna(0)
Output
fruit cost new
0 apples 2 0.0
1 oranges 3 0.0
2 bananas 5 0.0
3 grapefruit 1 0.0
0 apples 2 1.0
1 oranges 3 1.0
As an alternative you could reindex, with default_value=0, before concat:
filtered = df[df.fruit.isin(fruits)].assign(new=1)
result = pd.concat([df.reindex(columns=filtered.columns, fill_value=0), filtered], sort=False)
print(result)
Output
fruit cost new
0 apples 2 0
1 oranges 3 0
2 bananas 5 0
3 grapefruit 1 0
0 apples 2 1
1 oranges 3 1
df['indicator'] = df['fruit'].isin(fruits)?