[Python] Groupby a df column based on more than 3 columns

I have an df which has 3 columns lets say Region, Country and AREA_CODE.

Region Country AREA_CODE AREA_SUB_CODE_1 AREA_SUB_CODE_2
===========================================================================
AMER US A1 A1_US_1 A1_US_2
AMER CANADA A1 A1_CA_1 A1_CA_2
AMER US B1 B1_US_1 B1_US_2
AMER US A1 A1_US_1 A1_US_2


Is there a way to get output list of both the AREA_SUB_CODE_1 and AREA_SUB_CODE_2 as a list under each of the previous column value. something like the below

{
"AREA_SUB_CODE_1": {
"AMER": {
"US": {
"A1": ["A1_US_1"],
"B1": ["B1_US_1"]
},
"CANADA": {
"A1": ["A1_CA_1"],
}
}
},
"AREA_SUB_CODE_2": {
"AMER": {
"US": {
"A1": {
"A1_US_1": ["A1_US_2"]
},
"B1": {
"B1_US_1": ["B1_US_2"]
},
"CANADA": {
"A1": {
"A1_CA_1": ["A1_CA_2"],
}
}
}
},
}


So far i have tried to groupby on 3 columns it works which is,

for (k1, k2), v in df.groupby(['Region', 'Country'])['AREA_CODE']:
tTmp.setdefault(k1, {})[k2] = sorted(v.unique())



But when i try to groupby 4 columns, it is throwing error too many values to unpack (expected 2)

for (k1, k2), v in df.groupby(['Region', 'Country', 'AREA_CODE'])['AREA_SUB_CODE_1']:
tTmp.setdefault(k1, {})[k2] = sorted(v.unique())


Kindly suggest me a way on how to apply groupby for 4 columns and 5 columns. Or any other way to achieve this.

Thanks in advance.

Continue reading...