How to melt a dataframe so repeated items become the values that correspond to the index
Question:
I have this dataframe:
df = pd.DataFrame({'Status':['CO','AD','AD','AD','OT','CO','OT','AD'],
'Mutation':['H157Y','R47H','R47H','R67H','R62H','D87N','D39E','D39E']})
print(df)
Status Mutation
0 CO H157Y
1 AD R47H
2 AD R47H
3 AD R67H
4 OT R62H
5 CO D87N
6 OT D39E
7 AD D39E
I want the dataframe to look like this:
df2 = pd.DataFrame({'Status':['CO','AD','OT'],'H157Y':[1,0,0],'R47H':[0,2,0],'R67H':[0,1,0],
'R62H':[0,0,1],'D87N':[1,0,0],'D39E':[1,0,1]})
print(df2)
Status H157Y R47H R67H R62H D87N D39E
0 CO 1 0 0 0 1 1
1 AD 0 2 1 0 0 0
2 OT 0 0 0 1 0 1
Where mutations are the column names and their values – the number of hits – corresponds to the status.
Answers:
We can use pd.crosstab
like the below:
>>> pd.crosstab(df["Status"], df["Mutation"])
Mutation D39E D87N H157Y R47H R62H R67H
Status
AD 1 0 0 2 0 1
CO 0 1 1 0 0 0
OT 1 0 0 0 1 0
Or we can use pd.get_dummies
, pandas.DataFrame.groupby
then pandas.DataFrame.rename
columns like the below:
(pd.get_dummies(df,
columns=['Mutation']
).groupby(['Status']).sum().rename(columns=lambda x: x.split('_')[1]))
Output:
D39E D87N H157Y R47H R62H R67H
Status
AD 1 0 0 2 0 1
CO 0 1 1 0 0 0
OT 1 0 0 0 1 0
This should do the trick:
df.groupby(['Status', 'Mutation']).size().unstack(fill_value=0)
I have this dataframe:
df = pd.DataFrame({'Status':['CO','AD','AD','AD','OT','CO','OT','AD'],
'Mutation':['H157Y','R47H','R47H','R67H','R62H','D87N','D39E','D39E']})
print(df)
Status Mutation
0 CO H157Y
1 AD R47H
2 AD R47H
3 AD R67H
4 OT R62H
5 CO D87N
6 OT D39E
7 AD D39E
I want the dataframe to look like this:
df2 = pd.DataFrame({'Status':['CO','AD','OT'],'H157Y':[1,0,0],'R47H':[0,2,0],'R67H':[0,1,0],
'R62H':[0,0,1],'D87N':[1,0,0],'D39E':[1,0,1]})
print(df2)
Status H157Y R47H R67H R62H D87N D39E
0 CO 1 0 0 0 1 1
1 AD 0 2 1 0 0 0
2 OT 0 0 0 1 0 1
Where mutations are the column names and their values – the number of hits – corresponds to the status.
We can use pd.crosstab
like the below:
>>> pd.crosstab(df["Status"], df["Mutation"])
Mutation D39E D87N H157Y R47H R62H R67H
Status
AD 1 0 0 2 0 1
CO 0 1 1 0 0 0
OT 1 0 0 0 1 0
Or we can use pd.get_dummies
, pandas.DataFrame.groupby
then pandas.DataFrame.rename
columns like the below:
(pd.get_dummies(df,
columns=['Mutation']
).groupby(['Status']).sum().rename(columns=lambda x: x.split('_')[1]))
Output:
D39E D87N H157Y R47H R62H R67H
Status
AD 1 0 0 2 0 1
CO 0 1 1 0 0 0
OT 1 0 0 0 1 0
This should do the trick:
df.groupby(['Status', 'Mutation']).size().unstack(fill_value=0)