i want to split dataframe when value of flag column change to 1
df=pd.DataFrame({'A':[1,20,40,45,56,1,20,40,45,56],'flag':[3,2,4,1,1,3,3,1,1,1]})
Out[63]:
A flag
0 1 3
1 20 2
2 40 4
3 45 1
4 56 1
5 1 3
6 20 3
7 40 1
8 45 1
9 56 1
desired out:
print(group_1)
A flag
0 1 3
1 20 2
2 40 4
print(group_2)
A flag
0 1 3
1 20 3
You can use groupby on the masked DataFrame, with groups starting on each 1:
mask = df['flag'].eq(1)
groups = [g for _,g in df[~mask].groupby(mask.cumsum())]
output:
groups[0]
A flag
0 1 3
1 20 2
2 40 4
groups[1]
A flag
5 1 3
6 20 3
result = list()
flag = True
start = 0
for index,value in df['flag'].items():
if value==1 and index-1>0 and flag:
flag = False
result.append(df.iloc[start:index,])
elif value!=1 and not flag:
start = index
flag = True
