Find rows from a pandas DataFrame if there is another row matching some conditions in the data

Question

For example, if I have a DF like the following:

n from      km      to
0  B        300     A
1  A        300     B
2  D        290     A
3  B        310     C
4  A        290     D

I would like to select rows 0, 1, 2 and 4 since they all have another row in the same DF that has inverted from and to .

df2 = pd.DataFrame(columns=['to', 'from', 'km'])

for index, row in df.iterrows():
    f, t = row['from'], row['to']
    if ((df['to'] == f) & (df['from'] == t)).any():
        df2 = df2.append(row)

> df2

    to  from    km
0   A   B       300
1   B   A       300
2   A   D       290
4   D   A       290

Is it possible to do this without iteration over the rows?

Answer 1

Here is one way sort your columns and find the duplicated

s=pd.DataFrame(np.sort(df[['from','to']].values,1)).duplicated(keep=False)
yourdf=df[s]
yourdf
Out[32]: 
   n from   km to
0  0    B  300  A
1  1    A  300  B
2  2    D  290  A
4  4    A  290  D

Answer 2

Not as nice and short as the answer of Wen-Ben but maybe more intuitive. Merge the df with itself:

ok = df.merge(df[['from', 'to']], left_on='to', right_on='from').query('from_x == to_y')['n']
df.loc[ok, :]