CODEWITHSUNDEEP
pythonpandasdataframe
blakebullwinkel
blakebullwinkel

Reputation: 325

Pandas KeyError: 'occurred at index 0'

Let's say I have a Pandas dataframe df:

start_time   Event
0            0
1            0
2            0
3            0
4            0
5            0
6            0
7            0
8            0
9            0

I want to set the value of the Event column to -1 when the corresponding start_time lies between two values, so I define this function:

def test(time):
    if (time['start_time'] >= 5) and (time['start_time'] <= 8):
        return -1
    else:
        return time

To apply this to the event column, I do the following:

df[['Event']] = df[['Event']].apply(test,axis=1)

which yields this error: KeyError: ('start_time', 'occurred at index 0')

Why is this happening? Should be a simple fix.

Upvotes: 3

Views: 4383

Answers (2)

charlesreid1
charlesreid1

Reputation: 4821

The function that you are passing to .apply() uses the start_time field of the input argument (in the conditional checks if (time['start_time'] >= 5) and (time['start_time'] <= 8)). So it should be applied to a DataFrame or Series that has a start_time column.

However, before you call apply you are first calling df[['Event']], which returns a Series. So df[['Event']].apply() will apply a function to the resulting Series. But when the function reaches the expression time['start_time'], it is looking for a column called start_time in the Series, can't find it (because only 'Event' column was kept), and raises a KeyError.

The solution is to pass a DataFrame or a Series that has a start_time column in it. In your case you want to apply the function to the entire DataFrame so replace df[['Event']] with the whole DataFrame df.

df = df.apply(test, axis=1)

and change your function to modify the Event column instead of returning a value. Replace return -1 with time['Event'] = -1 and eliminate the else return time part (i.e., don't change anything if the conditions aren't met).

Upvotes: 1

ProteinGuy
ProteinGuy

Reputation: 1942

Simply do:

df['Event'] = df.apply(test, axis=1)['Event']

Upvotes: 2

Related Questions