Z-score calculation on rolling window

Question

How can I calculate z-score on 2 days rolling window with my data as below in pandas dataframe?

I also want to group by on class name.

Date	Class	Marks
01-01-2022	A	1700
02-01-2022	A	3000
03-01-2022	A	2624
04-01-2022	A	1745
05-01-2022	A	1789
06-01-2022	A	1874
01-01-2022	B	1965
02-01-2022	B	1847
03-01-2022	B	1849
04-01-2022	B	1754
05-01-2022	B	1598
06-01-2022	B	1515
01-01-2022	C	433
02-01-2022	C	350
03-01-2022	C	268
04-01-2022	C	433
05-01-2022	C	350
06-01-2022	C	268

I tried this.

my_data['zscore'] = (my_data['Marks']-(my_data['Marks'].rolling(2).mean()))/(my_data['Marks'].rolling(2).std())

Answer 1

Use:

def func(x):
    return np.mean(x)/np.std(x)

then:

my_data['Marks'].rolling(2).apply(func)

Or:

my_data['Marks'].rolling(2).apply(lambda x: np.mean(x)/np.std(x))

Based on your comment, use:

string = """Date    Class   Marks
01-01-2022  A   1700
02-01-2022  A   3000
03-01-2022  A   2624
04-01-2022  A   1745
05-01-2022  A   1789
06-01-2022  A   1874
01-01-2022  B   1965
02-01-2022  B   1847
03-01-2022  B   1849
04-01-2022  B   1754
05-01-2022  B   1598
06-01-2022  B   1515
01-01-2022  C   433
02-01-2022  C   350
03-01-2022  C   268
04-01-2022  C   433
05-01-2022  C   350
06-01-2022  C   268"""
temp = [x.split('   ') for x in string.split('\n')]
df=pd.DataFrame(temp[1:], columns = temp[0])
df.groupby('Class')['Marks'].apply(lambda x: np.mean(x)/np.std(x))

Output: