CODEWITHSUNDEEP
pythonparallel-processing
vgaurav
vgaurav

Reputation: 85

Extract return values of a function called in parallel

I would like to append the resulting dataframe i.e. the return dataframe from the createDF function to the full_df(an empty df).

In the run_parallel function below, I am passing args=(weight, eduList,) but I cannot seem to understand how to retrieve the return value from createDF and append it to full_df.

import pandas as pd 
import numpy as np
import random

# dummy function to create a df
def createDF(weight, eduList):
    #initialize list of lists 
    data = [['kappa',  weight*random.randint(5, 10), eduList[0]], 
            ['ombee',  weight*random.randint(5, 10), eduList[1]], 
            ['babad',  weight*random.randint(4, 10), eduList[2]], 
            ['matth',  weight*random.randint(4, 9),  eduList[3]], 
            ['allis',  weight*random.randint(4, 9),  eduList[4]]] 
    # Create the pandas DataFrame 
    df = pd.DataFrame(data, columns = ['Name', 'Score', 'Education']) 
    return df


weight = 9
depts = ['FIN', 'CONS', 'CONS', 'MANF', 'MED']
eduList = ['RWTH', 'EBS', 'OSU', 'OKSTATE', 'OK-State']

full_df = pd.DataFrame() # empty dataframe to which I would like to append all the return dfs from createDF

#function to create the dfs parallely
def run_parallel(weight, depts, eduList):
    from multiprocessing import Process, current_process
    processes = []

    for dept in depts:
        process = Process(target= createDF, args=(weight,eduList,))
        processes.append(process)
        process.start()

    for process in processes:
        process.join()



# desired result:
Name    Score   Education
0   kappa   72  RWTH
1   ombee   72  EBS
2   babad   40  OSU
3   matth   28  OKSTATE
4   allis   56  OK-State
5   kappa   63  RWTH
6   ombee   83  EBS
7   babad   60  OSU
8   matth   56  OKSTATE
9   allis   40  OK-State
..      ..      ..      ..
..      ..      ..      ..
..      ..      ..      ..

Upvotes: 1

Views: 339

Answers (1)

jonasfh
jonasfh

Reputation: 4569

You can define an empty list and append dataframes to this, and afterwards concat these to a single dataframe:

df_list = []
# dummy function to create a df
def createDF(weight, eduList):
    #initialize list of lists 
    data = [['kappa',  weight*random.randint(5, 10), eduList[0]], 
        ['ombee',  weight*random.randint(5, 10), eduList[1]], 
        ['babad',  weight*random.randint(4, 10), eduList[2]], 
        ['matth',  weight*random.randint(4, 9),  eduList[3]], 
        ['allis',  weight*random.randint(4, 9),  eduList[4]]] 
    # Create the pandas DataFrame 
    df = pd.DataFrame(data, columns = ['Name', 'Score', 'Education']) 
    df_list.append(df)
...
# And after all processes finish: 
full_df = pd.concat(df_list)

Upvotes: 1

Related Questions