ValueError：只能将大小为 1 的数组转换为 Python 标量

Question

I have a problem very similar to this one here:

Count all subordinates who directly or indirectly reports to managers

I do have a table with all employees in one column and their resp. boss_id in the other.

    employee_id boss_id receive_reports nr_reports
0   46456       175361  False           0
1   104708      29733   False           0
2   120853      41991   False           0
3   142630      171266  True            1
4   72711       198240  False           0

As a result I want a table, where for every employee I read the number of all direct and indirect reports (eg 0 for all people on the first level, up to #allemployees for the CEO). In the last columns I easily could differentiate between first level people and everybody else.

I want to use python and I've been wrapping my head around it for quite a while. I do have a general idea: starting with all the people at the first level and zero reports, going up by one level etc...

---

EDIT 1

here is my approach (I would be also curious to know if there is a better one)

I have a list with all level_0 employees.

def sum_reports(current_level):
   # pass current level of employees as list
   if len(current_level) == 1:
       return
   else:
       next_level = []
       #find boss for every employees at current level and add to next level + add to report of this boss 1
       for emp in current_level:
           tmp_boss = data.loc[data.employee_id == emp, "boss_id"].item()
           next_level.append(tmp_boss)
           data.loc[(data.employee_id == tmp_boss),"nr_reports"] += 1
       sum_reports(next_level)

   sum_reports(level_0)

But I do get the error:

ValueError: Can only compare identically-labeled Series objects

---

EDIT 2

Got rid of the comparison error in adding .item() changing the line of code with tmp_boss = ...

But now I do get another error, which I cannot track: ValueError: can only convert an array of size 1 to a Python scalar

Answer 1

I could fix the problem, doing the following:

• add a check, if a boss is already in my next level (as there are several reports to the same person)

• and check if the the boss_id is '0' - an entry I replaced the NaN for the CEO with. This actually caused the conversion error

  • to get along with imbalance in the tree (people directly reporting on higher level) I had to add another check - if there is a more concise or elegant solution, let me know.

Here is the final code:

#a list of the bosses column as the direct_reports
direct_reports = data.boss_id.tolist()
seen = []
def sum_reports(current_level):
    # pass current level of employees as list
    if len(current_level) <= 1:
        return
    else:
        next_level = []
        #find boss for every employee at current level and add to next level + add 1 to report of this boss
        for emp in current_level:
            tmp_boss = data.loc[data.employee_id == emp, "boss_id"].item()

            seen.append(tmp_boss)

            #only if is the last direct report: add boss to the next_level
            if direct_reports.count(tmp_boss) == seen.count(tmp_boss) and tmp_boss != 0:    
                next_level.append(tmp_boss)


            # add reports, plus 1 and the reports sofar of the current emp    
            sofar = data.loc[data.employee_id == emp, "nr_reports"].item()                
            data.loc[(data.employee_id == tmp_boss),"nr_reports"] += (sofar+1)    

        sum_reports(next_level)
# level_o is a list of all employees, who don't appear in the boss column
sum_reports(level_0)