pandas dataframe delete rows with low frequency

Question

What is the best practice to remove all rows that has a column with low frequency value?

Dataframe:

IN:
foo bar poo
1   a   A
2   a   A
3   a   B
4   b   B
5   b   A
6   b   A
7   c   C
8   d   B
9   e   B

Example 1: Remove all rows that have less than 3 in frequency value in column 'poo':

OUT:
foo bar poo
1   a   A
2   a   A
3   a   B
4   b   B
5   b   A
6   b   A
8   d   B
9   e   B

Example 2: Remove all rows that have less than 3 in frequency value in column 'bar':

OUT:
foo bar poo
1   a   A
2   a   A
3   a   B
4   b   B
5   b   A
6   b   A

Answer 1

This should generalise pretty easily. You'll need groupby + transform + count, and then filter the result:

col = 'poo'  # 'bar'
n = 3        # 2

df[df.groupby(col)[col].transform('count').ge(n)]

   foo bar poo
0    1   a   A
1    2   a   A
2    3   a   B
3    4   b   B
4    5   b   A
5    6   b   A
7    8   d   B
8    9   e   B

Answer 2

IIUC filter ..

df.groupby('poo').filter(lambda x : (x['poo'].count()>=3).any())
Out[81]: 
   foo bar poo
0    1   a   A
1    2   a   A
2    3   a   B
3    4   b   B
4    5   b   A
5    6   b   A
7    8   d   B
8    9   e   B

Or combine value_counts with isin

s=df.poo.value_counts().gt(3)
df.loc[df.poo.isin(s[s].index)]
Out[89]: 
   foo bar poo
0    1   a   A
1    2   a   A
2    3   a   B
3    4   b   B
4    5   b   A
5    6   b   A
7    8   d   B
8    9   e   B