How to fill empty cell value in pandas with condition - python-3.x

Question

My sample dataset is as below. Actuall data till 2020 is available.
Item Year Amount final_sales
A1 2016 123 400
A2 2016 23 40
A3 2016 6
A4 2016 10 100
A5 2016 5 200
A1 2017 123 400
A2 2017 23
A3 2017 6
A4 2017 10
A5 2017 5 200
I have to extrapolate 2017 (and subsequent years) final_sales column data from 2016 for every Item if 2017 data not available.
In the above dataset final_sales not available for the year 2017 for A2 and A4 but available for 2016 year. How to bring in 2016 data (final_sales) value if corresponding year final_sales not available?
Expected results as below. Thanks.
Item Year Amount final_sales
A1 2016 123 400
A2 2016 23 40
A3 2016 6
A4 2016 10 100
A5 2016 5 200
A1 2017 123 400
A2 2017 23 40
A3 2017 6
A4 2017 10 100
A5 2017 5 200

Answer 1

It looks like you want to fill forward where there is missing data.
You can do this with 'fillna', which is available on pd.DataFrame objects.
In your case, you only want to fill forward for each item, so first group by item, and then use fillna. The method 'pad' just carries forward in order (hence why we sort first).
df['final_sales'] = df.sort_values('Year').groupby('Item')['final_sales'].fillna(method='pad')
Note that on your example data, A3 is missing for 2016 as well, so there is nothing to carry forward and it remains missing for 2017.

Answer 2

For me working GroupBy.ffill, only necessary sorted Year column like in question sample data:
#if necessary sorting by both columns
df = df.sort_values(['Year', 'Item'])
df['final_sales'] = df.groupby('Item')['final_sales'].ffill()
print (df)
Item Year Amount final_sales
0 A1 2016 123 400.0
1 A2 2016 23 40.0
2 A3 2016 6 NaN
3 A4 2016 10 100.0
4 A5 2016 5 200.0
5 A1 2017 123 400.0
6 A2 2017 23 40.0
7 A3 2017 6 NaN
8 A4 2017 10 100.0
9 A5 2017 5 200.0

Answer 3

Something like this?:
def fill_final(x):
if x['year'] != 2016:
return df[(df['year'] == 2016) & (df['Item'] == x['Item'])]['final_sales']
else: return x['final_sales']
df['final_sales'] = df.apply(lambda x: fill_final(x), axis = 1)
did not test this but should set you on the right path