如何在我的 pandas apply 语句中避免此索引错误？

Question

I'm really lost on this error message, particularly as it seems to only occur in my 64 bit version of pandas, and not a slightly older 32 bit version I have in a different environment.

So I have two dataframes- "summary" and "all". "Summary" includes start times of types of events, whilst "all" contains data that's being interpreted into the summary table.

This line below that takes a 'starting reading' seems to be causing all the issues, not sure why though!

summary['average_after_start'] = summary.apply(lambda row: np.mean(all[(all['Timestamp'] >= row['Start'])]['Value']), axis=1, raw=True)

The error message I get is this:

IndexError: only integers, slices (`:`), ellipsis (`...`), numpy.newaxis (`None`) and integer or boolean arrays are valid indices

Any pointers would be appreciated!

*edit: I've simplified this a bit to be the simplest code I can find so far that causes the issue

Answer 1

If you set raw=True then row is a numpy array, not a pandas series and hence row['Start'] will throw an error as you can't index an array with a string.
So you should change it to raw=False (which is the default, so you can leave it out).