计算python Pandas中行之间的数据时间差异

Question

I'm trying to calculate the datetime difference between rows for each unique machine_id here. I have already grouped the Dataframes and have tried

newdf = newdf.copy()
newdf['diffs'] = float('nan')
newdf = newdf.copy()
for index in newdf.index.levels[0]:
    newdf.diffs[index] = newdf.event_datetime[index].diff

the dataset looks like

https://i.stack.imgur.com/eg93C.png

Answer 1

Have you tried diff after groupby operation? Something like:

newdf.groupby('machine_id').event_date.diff()

Answer 2

I tried to create multi index data frame, it should work fine using diff() function.

using newdf.groupby('machine_id').event_date.diff() suggested by ATL should work fine. o

# hierarchical indices and columns
index = pd.MultiIndex.from_product([[598, 615, 721], [43, 43, 45]],
                                   names=['machine_id', 'prod_category_id'])

# mock some data
data = ['2017-03-20 12:00:00','2017-03-29 01:00:00','2017-04-29 01:00:00',
        '2017-03-30 02:00:00', '2017-04-29 02:00:00','2017-05-29 12:00:00',
        '2017-10-30 02:00:00', '2017-11-29 02:00:00', '2017-11-29 04:00:00']

# create the DataFrame
newdf = pd.DataFrame(data, index=index)
newdf.columns = ['event_date']

newdf['event_date'] = pd.to_datetime(newdf['event_date'])
newdf.groupby(level=0)['event_date'].diff()