Reshifting Specific Column Based On Row If String Row Match With List

Okay, it's my real dataframe that i want to perform for reshaping on specific values with list. So i have this dataframe to reshape. [Out] = df Keterangan

Solution 1:

I see now what's causing the problem. If you look at the result of

match = df['Keterangan'].str.fullmatch('|'.join(entry for entry in my_list))
df['shift'] = match.cumsum()
df['index'] = df.index
columns = df.columns
df = df.apply(lambda row: print(row), axis='columns')

you will see something like

Q2 2019    22686796.0
Q2 2018    27421625.0
shift             0.0
index             0.0
Name: Kas, dtype: float64

Q2 2019    68409507.0
Q2 2018    71159442.0
shift             0.0
index             1.0
Name: Giro pada bank indonesia, dtype: float64

Q2 2019    15675129.0
Q2 2018    12584938.0
shift             1.0
index             2.0
Name: Giro pada bank lain, dtype: float64
...

The rows are Series with a unified type, which is float64 here.

From the documentation:

Objects passed to the function are Series objects ...

Some further points. If you look at the simple examples

df = pd.DataFrame({'A': [1, 2], 'B': [1., 2.]})
print(df.iloc[0], 'w')

df = pd.DataFrame({'A': ['a', 'b'], 'B': [1., 2.]})
print(df.iloc[0])

you will see the following output

A    1.0
B    1.0
Name: 0, dtype: float64

A    a
B    1
Name: 0, dtype: object

Both are Series. In the first, Pandas sees that all types are numeric, so it chooses the best numeric type to accomodate the types of both values, which is float. In the second, due to the string, Pandas chooses object, which accomodates almost all types.

In the DataFrame from your other question there's a string in the 2. and 3. column (the 'Nan', which isn't NaN!), which leads to the type object and in the following also to the type object in the row variables in apply. The DataFrame here has tpye float in the 2. and 3. column (the type of NaN is float) and therefore also float in the row variables. That's the reason why the original code worked for the first example but not here. (At least that's what I think, I could be wrong.)

I have adjusted the code in my suggestion accordingly (int casts).