CODEWITHSUNDEEP
pythonnumpypandas
SLE
SLE

Reputation: 85

Average of each day over multiple years

Quick Q:

Have a dataset: That is a measurment per day over several years. Trying to work out the average of all the same dates. i.e the average of 01/01/1995 and 01/01/1996 and 01/01/1997 etc.

Tried this: 

z=df.groupby(df.index.day,df.index.month).mean()

But get:

TypeError: unhashable type: 'numpy.ndarray'

Cheers

Upvotes: 3

Views: 2595

Answers (2)

EHB
EHB

Reputation: 1187

Hey @EdChum just adding this as an answer, as the display of the comment doesn't let me show what I meant. I agree that date format is a superior syntax! However, when I do it in reverse order I get

    WindSpeed

1   1   1   9.283333
1   2   6.694444
1   3   7.861111
1   4   9.223529
...
1   31  8.964706
2   1   9.127778
2   2   8.733333
2   3   7.511111
...

Where the first index is month, and the second is day. When I do in the order you suggested, I get:

    WindSpeed
1   1   9.283333
1   2   8.361111
1   3   6.670588
1   4   7.155556
1   5   8.450000
1   6   4.777778
1   7   3.666667
1   8   4.011765
1   9   5.210526
1   10  6.972222
1   11  7.647059
1   12  6.844444
...

Where the first index is day (from 1-31), and the second is month (from 1-12). We usually look at data sequentially, with January 1 followed by Jan 2. In yours, January first is followed by February first. That confused me for a bit! Mine worked better for my needs anyways. I appreciated your answer here a lot! Helped me with my problem.

Upvotes: 1

EdChum
EdChum

Reputation: 394409

IIUC you need to pass a list:

z=df.groupby([df.index.day,df.index.month]).mean()

What you did was pass multiple args so it interpreted the months array as an arg for axis, see the docs

Upvotes: 4

Related Questions