dask.compute Dask DataFrame 类型的所有值作为值存储在字典中

Question

I understand that if I store many Dask dataframes in a list I can compute all of them in parallel as

result = dask.compute(*container_list)

but how would I do something similar if I store the Dask dataframe results as values in a dictionary? (If containe_dict is a dictionary

result = dask.compute(*container_dict) 

would not work.)

The best I could do was loop over the dictionary with a container, but this is not ideal since we are now running dask.compute multiple times rather than once.

container_dict = {}
for index, value in enumerate(comb_dict_stock):
    container_dict[index] = ddf.loc[index] # index ddf to get the row for index and value in dict

# compute all the dask dataframes in container_dict
for key, value in container_dict.items():
    container_dict[key] = value.compute()

Answer 1

dask.compute can accept a dictionary and evaluate only the dask objects inside:

from dask import compute
from dask.datasets import timeseries
test = {'a': timeseries(freq='1h'), 'b': 123}
result, = compute(test)
print(type(result))
# <class 'dict'>

Note that compute returns a tuple of results, so to store just the dictionary of interest use tuple assignment.