Pandas Resampling: TypeError: Only valid with DatetimeIndex, TimedeltaIndex or PeriodIndex, but got an instance of 'RangeIndex'

Question

Please, help me. I want to resample based on 1D. I have following format of data. I want to use resampling in pandas.

I want to resample based on Date and product and also fill the missing values.

But I keep getting this mistake: I tried like 5 options and mistake only changes after "instance of": I saw there Multiindex, Index.

TypeError: Only valid with DatetimeIndex, TimedeltaIndex or PeriodIndex, but got an instance of 'RangeIndex'

product value   date
A   1.52    2016-01-01
A   NULL    2016-09-20
A   1.33    2018-08-02
B   1.30    2016-01-01
B   NULL    2017-01-02
B   1.54    2017-03-10
B   2.08    2017-06-28
B   2.33    2018-08-02

I put these data into

df.reset_index().set_index('date','sku')  
df= df.groupby('product').resample('1D')['value'].ffill().bfill().ffill()

I tried also:

df = df.set_index(['date','sku'])
df = df.set_index('date','sku')
df = df.reset_index().set_index(['date','sku'])  

Please, can you explain me what I am doing wrong? Thanks!

Today morning it was working on these data and the command from Jezrael:

df = df.set_index('date').groupby('product').resample('1D')['value'].ffill()

    product value   date
   0    A   1.52    2016-01-01
   1    A   NaN 2016-09-20 
   2    A   1.87    2018-08-02
   3    B   2.33    2016-01-01
   4    B   NaN 2016-09-20
   5    B   4.55    2018-08-02

But suddenly it doesnt anymore. Now I have Index in the error statement.

Answer 1

You need DatetimeIndex if working with DataFrameGroupBy.resample, also bfill is omited because if some only NaNs groups is possible these data are replaced from another groups:

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

df = df.set_index('date').groupby('product').resample('1D')['value'].ffill()
print (df)
product  date      
A        2016-01-01    1.52
         2016-01-02    1.52
         2016-01-03    1.52
         2016-01-04    1.52
         2016-01-05    1.52
         2016-01-06    1.52
         2016-01-07    1.52
         2016-01-08    1.52
         2016-01-09    1.52
         2016-01-10    1.52
         2016-01-11    1.52
         2016-01-12    1.52

Changed sample for better explanation:

print (df)
  product  value       date
0       A   1.52 2016-01-01
1       A    NaN 2016-01-03
2       B    NaN 2017-01-02
3       B    NaN 2017-01-03
4       C   1.54 2017-03-10
5       C   2.08 2017-03-12
6       C   2.33 2017-03-14

---

df1 = df.set_index('date').groupby('product').resample('1D')['value'].ffill()
print (df1)
product  date      
A        2016-01-01    1.52
         2016-01-02    1.52
         2016-01-03     NaN < NaN is not changed because in original data
B        2017-01-02     NaN <- only NaN group B
         2017-01-03     NaN
C        2017-03-10    1.54
         2017-03-11    1.54
         2017-03-12    2.08
         2017-03-13    2.08
         2017-03-14    2.33
Name: value, dtype: float64

df11 = df.set_index('date').groupby('product').resample('1D')['value'].ffill().bfill()
print (df11)
product  date      
A        2016-01-01    1.52
         2016-01-02    1.52
         2016-01-03    1.54 <- back filling value from group C
B        2017-01-02    1.54 <- back filling value from group C
         2017-01-03    1.54 <- back filling value from group C
C        2017-03-10    1.54
         2017-03-11    1.54
         2017-03-12    2.08
         2017-03-13    2.08
         2017-03-14    2.33
Name: value, dtype: float64