CODEWITHSUNDEEP
pythonpandas
FunnyChef
FunnyChef

Reputation: 1946

How to count only specific values in a pandas dataframe

I have the following pandas dataframe;

a = [['01', '12345', 'null'], ['02', '78910', '9870'], ['01', '23456', 'null'],['01', '98765', '8760']]

df_a = pd.DataFrame(a, columns=['id', 'order', 'location'])

I need to get a count of how many NULL values (NULL is a string) that occur for each ID. So the result would look like;

id   null_count
01    02

I can get basic counts using a groupby:

new_df = df_a.groupby(['id', 'location'])['id'].count()

But the results return more than just the NULL values;

id  location
01  8760        1
    null        2
02  9870        1

Upvotes: 4

Views: 114

Answers (3)

BENY
BENY

Reputation: 323226

Base on your own code , adding .loc notice this is multi index slice ..

df_a.groupby(['id', 'location'])['id'].count().loc[:,'null']
Out[932]: 
id
01    2
Name: id, dtype: int64

Upvotes: 5

MaxU - stand with Ukraine
MaxU - stand with Ukraine

Reputation: 210832

In [16]: df_a.set_index('id')['location'].eq('null').sum(level=0)
Out[16]:
id
01    2.0
02    0.0
Name: location, dtype: float64

Upvotes: 4

Scott Boston
Scott Boston

Reputation: 153460

Because in your source dataframe your NULLs are strings 'null', use:

df_a.groupby('id')['location'].apply(lambda x: (x=='null').sum())\
    .reset_index(name='null_count')

Output:

   id  null_count
0  01          2
1  02          0

OR

df_a.query('location == "null"').groupby('id')['location'].size()\
    .reset_index(name='null_count')

Output:

   id  null_count
0  01           2

Upvotes: 6

Related Questions