For each group in Pandas dataframe, return the most common value if it shows up more than `x%` of the time

Question

Given a pandas dataframe, I would like to return a column's (string datatype) most common value for each groupby if this value shows up in more than n% of the rows, otherwise return 'NA'.

Answer 1

If need test number of most common values by count:

N = 5
def f(x):
    y = x.value_counts()
    return y.index[0] if y.iat[0] > N else np.nan


df = df.groupby('g')['col'].agg(f).reset_index(name='new')

Or by percentages:

n = 50
def f(x):
    y = x.value_counts(normalize=True) * 100
    return y.index[0] if y.iat[0] > n else np.nan


df = df.groupby('g')['col'].agg(f).reset_index(name='new')