根据 pandas 数据框中另一列中的条件对一列求和

Question

I just started using Python and I am trying to create programs to help monitor some of my investments. Right now I have a definition set up that will give me my current returns based on my initial buy price and the current price. Here is what my data frame looks like:

    Ticker   Expiration    Contracts   Call   Buy   Prem 12/22  Prem 12/23  Prem 12/25
0   x          date 1        1         $      0.13   0.15         0.12       0.13
1   y          date 2        1         $      0.33   0.34         0.34       0.39
2   z          date 3        1         $      0.25   NaN          NaN        0.25

I have the current definition written for returns:

def returns(op):
    """
    Calculates the current return for each options
    """
    totalPrem=op.sum(axis=0,skipna=True)["Prem 12/22":]
    buy=op.sum(axis=0,skipna=True)["Buy"]
    return (totalPrem-buy)*100

This gives me the results by adding all the columns from Prem 12/22 onward and subtracting it from the sum of the Buy column. My problem is that on 12/22 and 12/23, z was not yet bought. However, the returns definition sums all of Buy. So the returns for 12/22 and 12/23 adds the two data points in 12/22 and 12/23 and subtracts them from the 3 data points in Buy. This leads to the result:

Prem 12/22: -22
Prem 12/23: -25
Prem 12/25: 6

I want to alter my code to where for 12/22 and 12/23, the buy column only adds the first two. I was wondering if there was a way to where buy could be calculated by summing the buy column in a way where the data points are only added together if there is no NaN on the row of the data point. The output I am looking for is:

Prem 12/22: 3
Prem 12/23: 0
Prem 12/25: 6

Thanks!

Answer 1

You can use list comprehension to filter for notnull() rows by column and do the calculation per column. To only apply to the columns with Prem in them, I create a cols index object so we can dynamically apply changes to those indexed columns:

cols = df.columns[df.columns.str.contains('Prem')]
res = [int(round((df.loc[df[col].notnull(), col].sum() -
                  df.loc[df[col].notnull(), 'Buy'].sum()), 3) 
                  * 100)  for col in cols]
for c,r in zip(cols, res):
    print(f'{c}: {r}')

Prem 12/22: 3
Prem 12/23: 0
Prem 12/25: 6

Answer 2

Your purpose is to calculate return of Buy column. Suppose your data frame that contain Buy column is called df then you can simply call like the following:

def returnBuy(df):
    tsBuy=df["Buy"]
    return tsBuy.pct_change()

Answer 3

May be this: you could create a boolean column: if NA/NAN After that take the rows without missing data and sum after that:

data = {'col_1': [1, 2, 3, 4], 'col_2': [5, 7, pd.NA, 8]}
df = pd.DataFrame.from_dict(data)
df['should_add'] = pd.isna(df["col_2"])
print(df)
sums= df[~df['should_add']].sum(axis=0)
print(sums)

or one line:

sums= df[~pd.isna(df["col_2"])].sum(axis=0)