Python Pandas Dataframe 用列表中的值替换 NaN

Question

I'm trying to replace my column with NaN

group_choices = ['Group1', 'Group2', 'Group3']

Groups limit
1 NaN NaN
2 Group1 2
3 Group2 2
4 Group3 2
5 NaN NaN
6 NaN NaN
7 NaN NaN

How can I replace NaN, randomly based on the group_choises?

I'm also trying to limit how often a group_choise can be randomised selected because of the limit value in the limit column.

I'm trying to get this result:

Groups limit
1 Group3 NaN
2 Group1 2
3 Group2 2
4 Group3 2
5 Group1 NaN
6 Group2 NaN
7 Out of groups

Answer 1

fillna with dictionaries

dct = dict(zip(df.Groups.loc[pd.isna].index, group_choices))

df.fillna({'Groups': dct}).fillna({'Groups': 'Out of groups'})

          Groups  limit
1         Group1    NaN
2         Group1    2.0
3         Group2    2.0
4         Group3    2.0
5         Group2    NaN
6         Group3    NaN
7  Out of groups    NaN

---

Old Answers

Useful but I like the new one better. It illustrates the evolution of my thought process.

Generators

def get_some(i, n):
  for x in [*i] * n:
    yield x

def fill(s, i, n):
  gs = get_some(i, n)
  for x in s:
    if pd.isnull(x):
      try:
        yield next(gs)
      except StopIteration:
        yield "Out of groups"
    else:
      yield x

df.assign(Groups=[*fill(df.Groups, group_choices, 1)])

          Groups  limit
1         Group1    NaN
2         Group1    2.0
3         Group2    2.0
4         Group3    2.0
5         Group2    NaN
6         Group3    NaN
7  Out of groups    NaN

---

Alternative

def get_some(i, n):
  for x in [*i] * n:
    yield x

df.assign(Groups=df.Groups.fillna(
    df.Groups.loc[pd.isna].pipe(
        lambda s: pd.Series(dict(zip(s.index, get_some(group_choices, 1))))
    )
).fillna('Out of groups'))