How to groupby external time series data together - pandas

Question

I have been trying to groupby some state data together. This is how my data looks like for example, with Date as the index and the rest is features:
Date
Population
Num_Men
Num_Women
State
Region
2020-01-01
500
300
200
NY
North
2020-02-01
800
500
300
GL
Middle
2020-02-01
1000
400
600
""
Middle
2020-02-01
200
50
150
nan
Middle
2020-02-01
600
400
200
NY
North
I know how to group the NY states ones, but if I want to group the ones with state values: GL, "", and nan together. I'm not sure how to do that.
I was looking for the final result to look like:
Date
Population
Num_Men
Num_Women
State
Region
2020-01-01
500
300
200
NY
North
2020-02-01
2000
950
1050
GL
Middle
2020-02-01
600
400
200
NY
North
I did something like this: df.groupby(df.index, {'State': ["GL", "", np.nan]}, but that didn't work. Any help would be appreciated! Thanks!

Answer 1

Let us do replace then groupby with sum and first
df.State = df.State.replace({"''":np.nan,'nan':np.nan})
out = df.groupby(['Region','Date'],as_index=False).\
agg({'Population':'sum',
'Num_Men':'sum',
'Num_Women':'sum',
'State':'first'})
Out[99]:
Region Date Population Num_Men Num_Women State
0 Middle 2020-02-01 2000 950 1050 GL
1 North 2020-01-01 500 300 200 NY
2 North 2020-02-01 600 400 200 NY