多个数据帧之间如何关联？ Python Pandas

Question

I have multiple df like this:

df1:

date_time               Value(0 or 1)
2020-08-28 07:33:52     0
2020-08-28 08:08:51     0
2020-08-28 08:31:31     0
2020-08-28 08:31:59     0
2020-08-28 08:34:44     0
2020-12-24 10:10:08     1


df2:

date_time             rpm
2020-08-27 20:42:02   0.000000
2020-08-28 07:31:12   0.000000
2020-08-28 07:33:04   0.000000
2020-08-28 08:28:53   -0.001589
2020-08-28 08:29:51   -0.001589
2020-08-28 18:21:42   104.971931


df3:

date_time               Step
2020-08-28 07:33:52     1
2020-08-28 08:08:51     5
2020-08-28 08:31:59     10
2020-08-28 08:34:44     15
2020-08-28 08:36:26     20
2020-12-07 16:49:22     25

I would like to study the correlation between this dataframes, but I have a technical question, do I have to merge the dataframes and do correlation between columns? or there is an other way? and how do to it?

As you can see the seconds columns for each df are completely differents (others units).

Answer 1

If you want to merge to the nearest timestamp

df1['date_time'] = pd.to_datetime(df1['date_time'])
df2['date_time'] = pd.to_datetime(df2['date_time'])
df3['date_time'] = pd.to_datetime(df3['date_time'])

out = pd.merge_asof(df1, df2, on='date_time')
out = pd.merge_asof(out, df3, on='date_time')
out.corr()

If you want to ignore the timestamps and instead simply concatenate

pd.concat([df1,df2['rpm'],df3['Step']], axis=1).corr()

These will yield different results, because merge_asof is looking at the timestamps and merging the values together on the closest timestamp from df1