Dask error when compute() is called more that one time

Question

I'm working with dask and I need to perform a transformation using compute() before to continue the workflow. During the second transformation seems that the object change of type and is raising the error 'DataFrame' object has no attribute 'compute'

My current code looks like:

newtypes = {"age":int, "is_superuser":bool}
df = df.astype(newtypes).compute()
# ....
# 
....
strtypes = {"age":str, "is_superuser":str)
df = df.astype(strtypes).compute() # Here is raised the exception.

Check ``type(df)`` it might be pandas dataframe which does not support ``compute`` — sushanth, Jun 26 '20 at 03:29
You're right, it change from `` to ``. Any idea of how to solve it ? I read about `delayed` but since I need to perform the transformation immediately it's not my option. — Carlos Rojas, Jun 26 '20 at 03:40
This might be helpful, https://stackoverflow.com/a/39722445/4985099 — sushanth, Jun 26 '20 at 03:42
Please first consider what it is you are trying to achieve, and rephrase your question. — mdurant, Jun 26 '20 at 14:57

score 0 · Answer 1 · answered Jun 27 '20 at 16:36

0

As mentioned in the comments, your second dataframe is actually a Pandas dataframe, not a Dask dataframe, and so you do not need to call .compute() the second time.

answered Jun 27 '20 at 16:36

MRocklin

55,641
23
163
235

Dask error when compute() is called more that one time

1 Answers1