基于一个值添加新列

Question

I don't know if this is the right title for my question. Anyway, this is my simplified df:

    change   gain
0      NaN    NaN
1    216.0  216.0
2    270.0  270.0
3    167.0  167.0
4    -80.0    0.0
5   -338.0    0.0
6   -155.0    0.0
7    253.0  253.0
8   -385.0    0.0
9    -30.0    0.0
10   198.0  198.0
11   318.0  318.0
12   266.0  266.0
13   142.0  142.0
14    85.0   85.0
15   199.0  199.0
16    53.0   53.0
17    91.0   91.0
18   187.0  187.0
19    64.0   64.0
20   118.0  118.0

I want to add another column called average_gain that starts at row 14 and that row value is equal to: df['gain'].head(14).mean() Which is 140.769231. So far my desired output looks like this:

   change   gain  average_gain
0      NaN    NaN           NaN
1    216.0  216.0           NaN
2    270.0  270.0           NaN
3    167.0  167.0           NaN
4    -80.0    0.0           NaN
5   -338.0    0.0           NaN
6   -155.0    0.0           NaN
7    253.0  253.0           NaN
8   -385.0    0.0           NaN
9    -30.0    0.0           NaN
10   198.0  198.0           NaN
11   318.0  318.0           NaN
12   266.0  266.0           NaN
13   142.0  142.0           NaN
14    85.0   85.0    140.769231
15   199.0  199.0           NaN
16    53.0   53.0           NaN
17    91.0   91.0           NaN
18   187.0  187.0           NaN
19    64.0   64.0           NaN
20   118.0  118.0           NaN

Then for rows that are below row 14, I want to use this code,for example for row 15:

df.loc[15, 'average_gain'] = (df.loc[14, 'average_gain']*13)+df.loc[15,'gain'] It is the same formula for the rest of the rows. So if the row number is n, the formula is:

df.loc[n, 'average_gain'] = (df.loc[n-1, 'average_gain']*13)+df.loc[n,'gain']

I tried these two ways but didn't work:

def func(x):
    df['average_gain'].loc[n] = (df['average_gain'].loc[n - 1] * 13) + df['gain'].loc[n]
    return x

df['average_gain'].loc[15:] = (df['average_gain'].shift(1) * 13) + df['gain']

Answer 1

you can use:

df['average_gain'] = np.NaN
df['average_gain'][14] =  df['gain'].head(14).mean()
df

then apply your formula:

for i in range(15, len(df)):
    df.loc[i, 'average_gain'] = df.loc[i-1, 'average_gain'] * 13 + df.loc[i,'gain']
df