遍历熊猫行并根据其他列中的值设置列值

Question

I have a dataframe, one column (col1) of which contains values either Y or N. I would like to assign values (random, not repetitive numbers) to the next column (col2) based on the values in col1 - if value in col1 equals to N, then value in col2 would be some number, if value in col1 equals to Y, then value in col2 would repeat the previous. I tried to create a for loop and iterate over rows using df.iterrows(), however the numbers in col2 were equal for all Ns.

Example of the dataframe I want to get:

df = pd.DataFrame([[N, Y, Y, N, N, Y], [1, 1, 1, 2, 3, 3]])

where for each new N new number is assigned in other column, while for each Y the number is repeated as in previous row.

Answer 1

Assuming a DataFrame df:

df = pd.DataFrame(['N', 'Y', 'Y', 'N', 'N', 'Y'], columns=['YN'])
    YN
0   N
1   Y
2   Y
3   N
4   N
5   Y

---

Using itertuples (no repeation):

np.random.seed(42)
arr = np.arange(1, len(df[df.YN == 'N']) + 1)
np.random.shuffle(arr)

cnt = 0
for idx, val in enumerate(df.itertuples()):
    if df.YN[idx] == 'N':
        df.loc[idx, 'new'] = arr[cnt]
        cnt += 1
    else:
        df.loc[idx, 'new'] = np.NaN
df.new = df.new.ffill().astype(int)
df
    YN  new
0   N   1
1   Y   1
2   Y   1
3   N   2
4   N   3
5   Y   3

---

Using apply (repetition may arise with small number range):

np.random.seed(42)
df['new'] = df.YN.apply(lambda x: np.random.randint(10) if x == 'N' else np.NaN).ffill().astype(int)
    YN  new
0   N   6
1   Y   6
2   Y   6
3   N   3
4   N   7
5   Y   7