pandas create new dataframe based on information of each row on a different dataframe

Question

This is what I have:

ID  PRICE   VOLUME  PRODUC      FROM_DATE   TO_DATE         NUMDAYS

1   20.5    15.0    prod_1      2018-08-06      2018-08-13      7
2   15.6    10.0    prod_2      2018-08-06      2018-08-08      2

This is what I want to achieve:

ID  PRICE   VOLUME  PRODUC      FROM_DATE   TO_DATE         NUMDAYS

1   20.5    15.0    prod_1      2018-08-06      2018-08-07      1
1   20.5    15.0    prod_1      2018-08-07      2018-08-08      1
1   20.5    15.0    prod_1      2018-08-08      2018-08-09      1
1   20.5    15.0    prod_1      2018-08-09      2018-08-10      1
1   20.5    15.0    prod_1      2018-08-10      2018-08-11      1
1   20.5    15.0    prod_1      2018-08-11      2018-08-12      1
1   20.5    15.0    prod_1      2018-08-12      2018-08-13      1
2   15.6    10.0    prod_2      2018-08-06      2018-08-07      1
2   15.6    10.0    prod_2      2018-08-07      2018-08-08      1

So I have a Dataframe with information about products that affect different dates.

• Products may affect from 1 day to n days.
• The volume affects each date in between.

How could I do it?

I have tryed: - To do a for loop for each element of the dataframe but

df_results = pd.DataFrame(columns=df.columns)
for index, row in df.iterrows():
    day = row.to_dict()
    for i in range(0,int(row['numdays'])):
        day['NUMDAYS'] = 1
        day['FROM_DATE'] = row['FROM_DATE']+datetime.timedelta(days=i)
        day['TO_DATE'] =  day['FROM_DATE'] + datetime.timedelta(days=1)
        df_aux = pd.DataFrame.from_dict(day)
        df_results .append(df_aux)

However I can't make it work.

Answer 1

In pandas is best avoid loops, because slow:

#convert columns to datetimes if necessary
df['FROM_DATE'] = pd.to_datetime(df['FROM_DATE'])
df['TO_DATE'] = pd.to_datetime(df['TO_DATE'])

#repeat rows
df = df.loc[np.repeat(df.index, df['NUMDAYS'])]

#add timedeltas by counter
df['FROM_DATE'] += pd.to_timedelta(df.groupby('ID').cumcount(), unit='d')
#add one dau
df['TO_DATE'] = df['FROM_DATE'] + pd.Timedelta(1, unit='d')
#assign scalar
df['NUMDAYS'] = 1
#create default unique index
df = df.reset_index(drop=True)
print (df)
   ID  PRICE  VOLUME  PRODUC  FROM_DATE    TO_DATE  NUMDAYS
0   1   20.5    15.0  prod_1 2018-08-06 2018-08-07        1
1   1   20.5    15.0  prod_1 2018-08-07 2018-08-08        1
2   1   20.5    15.0  prod_1 2018-08-08 2018-08-09        1
3   1   20.5    15.0  prod_1 2018-08-09 2018-08-10        1
4   1   20.5    15.0  prod_1 2018-08-10 2018-08-11        1
5   1   20.5    15.0  prod_1 2018-08-11 2018-08-12        1
6   1   20.5    15.0  prod_1 2018-08-12 2018-08-13        1
7   2   15.6    10.0  prod_2 2018-08-06 2018-08-07        1
8   2   15.6    10.0  prod_2 2018-08-07 2018-08-08        1