my main goal is to apply a function on a dataframe rows and add columns to that dataframe containing the results of each row. example :

to this result :

Where avg_inv,proc and shortage are outputs of this function:

def order_up_to(demand_df,info_df,run,bk1,bk2):
...........
return proc,avg_inv,shortage

this function works well when i try it with one (run,bk1,bk2) it returns 3 integers but when i try to loop through all the rows and get the results in different lists i get this error :

  avg=[]
proc=[]
shor=[]
for index,row in slc.iterrows():
    avg.append(order_up_to(dmd,df,row["run_id"],row["bk1"],row["bk2"])[0])
    proc.append(order_up_to(dmd,df,row["run_id"],row["bk1"],row["bk2"])[1])
    shor.append(order_up_to(dmd,df,row["run_id"],row["bk1"],row["bk2"])[2])
---------------------------------------------------------------------------
TypeError                                 Traceback (most recent call last)
Input In [126], in <cell line: 4>()
      3 shor=[]
      4 for index,row in slc.iterrows():
----> 5     avg.append(order_up_to(dmd,df,row["run_id"],row["bk1"],row["bk2"])[0])
      6     proc.append(order_up_to(dmd,df,row["run_id"],row["bk1"],row["bk2"])[1])
      7     shor.append(order_up_to(dmd,df,row["run_id"],row["bk1"],row["bk2"])[2])

Input In [99], in ou2(demand_df, info_df, run, bk1, bk2)
     12 purchase_qty=[0]*len(demand)
     14 ######policy logic
---> 15 ending_inv_lvl[0]=initial_inv-demand[0]
     16 if ending_inv_lvl[0]<=safety_stock :
     17     purchase_qty[0]=order_qty-ending_inv_lvl[0]

TypeError: unsupported operand type(s) for -: 'NoneType' and 'int'

CodePudding user response：

applying the function to the dataframe columns instead of iterating through the rows makes it much easier:

def order_up_to(run_id, bk1, bk2):
    # your function logic here
    return avg_inv, proc, shortage

df = pd.DataFrame({'run_id': [2,4], 'bk1': [1000, 1890], 'bk2': [3, 3]})
[df["avg_inv"], df["proc"], df["shortage"]] = order_up_to(df.run_id, df.bk1, df.bk2)

CodePudding user response：

Rather than looping through the DataFrame, you could use the apply function and then split the result into separate columns -

d = dict(zip(['run_id', 'bk1', 'bk2'], [[2, 4], [1000, 1890], [3, 3]]))
df = pd.DataFrame(d)

def tuple_returning_func(row):
    return 2*row['run_id'], 3*row['bk1'], 4*row['bk2']

df[list('abc')] = df.apply(tuple_returning_func, axis=1).apply(pd.Series)

output

   run_id   bk1  bk2  a     b   c
0       2  1000    3  4  3000  12
1       4  1890    3  8  5670  12