Aggregates values for rows with duplicate index keys

aggretate_dups_csv.py

import pandas

value_keys = [
    "MONTO_APROBADO",
    "MONTO_MODIFICADO",
    "MONTO_EJERCIDO"
]

FILENAME = 'OS_Presupuesto_2018CDMX'

# Make sure OS_Presupuesto_2018CDMX.csv is saved as utf-8.
# Assign str type to all columns
df = pandas.read_csv('{}.csv'.format(FILENAME), encoding='utf-8', dtype=str)

# Assign float to value columns
for k in value_keys:
    df[k] = df[k].astype(float)

index_keys = [k for k in df.keys().values if k not in value_keys]

# We have to fill empty cells, otherwise they are NaN which will be ignored by
# the df.groupby.
df = df.fillna('')

# Retain first duplicate row, then drop the rest.
# df.drop_duplicates(inplace=True, subset=index_keys)

# Or, separately sum value columns for duplicates.
df = df.groupby(index_keys, as_index=False)[value_keys].agg('sum').round(2)

df.to_csv('{}.aggduped.csv'.format(FILENAME), index=False)