如何使用 python pandas 根据条件创建“计数”列？

Question

I have used groupby like below but it didn't work:

df={ 'id' :[1,1, 2,2, 3], 'testname' : ['math', 'science', 'math', 'literature', 'math'], 'result' :['passed', 'failed', 'passed', 'passed', 'failed'}


    ndf=df.groupby(['id', 'testname']) ['result']. count() 

Example dataframe:

Id  testname.    result
1.     math.           passed
1.     science.      failed 
2.     math.           passed
2.     literature.    passed
3.     math.           failed

Based on condition: count+=1 if the id is pass all exam that he take else count =0.

Therefore, the output should be like:

Expected output: Get a total value - >Total pass student will be 1.

Answer 1

It seems you are looking to count the number of students who don't have any failed tests (passed everything). You are on the right track with grouping...but I'm not sure why you are grouping by id and testname .

Sometimes in these types of problems where you are looking for "the ones that don't have any negative results" you can more easily count the ones with any negative result and subtract that from the original dataset size. Here is an approach:

1. find any row with a failure using filtering
2. group that result by id so now we have a basis to count everybody with a failure
3. subtract that count from the original number of unique id's

Note: You could certainly chain some of this stuff together, I just broke it apart for clarity.

In [25]: df                                                                     
Out[25]: 
   id    testname  result
0   1        math  passed
1   1     science  failed
2   2        math  passed
3   2  literature  passed
4   3        math  failed

In [26]: failed_df = df[df['result']=='failed']                                 

In [27]: ids_with_failures = len(failed_df)                                     

In [28]: tot_ids = len(df.groupby('id'))                                        

In [29]: count = tot_ids - ids_with_failures                                    

In [30]: count                                                                  
Out[30]: 1

Answer 2

df = pd.DataFrame({'id': [1, 1, 2, 2, 3],
                   'testname': ['math', 'science', 'math', 'literature', 'math'],
                   'result': ['passed', 'failed', 'passed', 'passed', 'failed']})

This function below returns True if all entries in a series are equal to 'passed'. In other words if the student has not 'passed' at least once, it returns False.

def verify_all_exams_are_passed(results):
    return results.eq('passed').all()

Finally, apply the function to each student (id), to see which students have passed all exams.

>>> students_passed = df.groupby('id')['result'].apply(verify_all_exams_are_passed)

>>> students_passed
id
1    False
2     True
3    False
Name: result, dtype: bool

You can also sum this series directly to get the count of students who passed all exams

>>> total_passed.sum()
1