值未在数据框中返回

Question

I am trying to drop rows with a 'count' values of less than 10 in my dataframe. My dataframe currently looks something like this:

    person  id     count
0   p1  760431192   20
1   p2  101663519   1
2   p3  325694288   2
3   p4  338468584   1
4   p5  2337087786  18

I merged the count column with the df.merge function based off of the id column:

df = df.merge(dframe, on='id', how='left')

So when I try to drop rows with a count < 10, i get the following error:

df = df[df.count>=10]
KeyError: True

However, when I use this same code on any other column, say:

df = df[df.id==760431192]
df = df[df.person==p2]

The code works perfectly, and i get the dataframe I was expecting. Any idea why the code is not working on the merged column 'count'?

Answer 1

df.count isn't the column, it's the method DataFrame.count . So you're not comparing a dataframe against a number (giving elementwise boolean results), you're comparing a method against a number, which there's no rule for. In Python 2, when there's no rule for a comparison, it falls back to a default "arbitrary but consistent" rule, which gives a single boolean answer.

In Python 3, that default rule has been removed, and the error you get gives you a much better idea of what's going on:

>>> df.count >= 10
Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
TypeError: unorderable types: method() >= int()

In any case, the solution is to get that column as df['count'] instead:

>>> df[df['count'] >= 10]
  person          id  count
0     p1   760431192     20
4     p5  2337087786     18

Answer 2

Another way to add the count of unique items back to the original DataFrame is to use groupby together with transform :

df['count'] = df.groupby('id').transform('count')

You can now filter out the rows with a count less than ten:

df = df[df['count'] >= 10]