CODEWITHSUNDEEP
python-3.xpandaspandas-groupby
MBasith
MBasith

Reputation: 1499

Pandas Group-By and Calculate Ratio of Two Columns

I'm trying to use Pandas and groupby to calculate the ratio of two columns. In the example below I want to calculate the ratio of staff Status per Department (Number of Status in Department/Total Number of Employees per Department). For example the Sales department has a total of 3 Employees and the number of staff that have Employee Status is 2 which gives the ratio of 2/3, 66.67%. I managed to hack my way through to get this but there must be a more elegant and simple way to do this. How can I get the desired output below more efficiently?

Original DataFrame:

  Department    Name      Status
0      Sales    John    Employee
1      Sales   Steve    Employee
2      Sales    Sara  Contractor
3    Finance   Allen  Contractor
4  Marketing  Robert    Employee
5  Marketing    Lacy  Contractor

Code:

mydict ={
        'Name': ['John', 'Steve', 'Sara', 'Allen', 'Robert', 'Lacy'],
        'Department': ['Sales', 'Sales', 'Sales', 'Finance', 'Marketing', 'Marketing'],
        'Status': ['Employee', 'Employee', 'Contractor', 'Contractor', 'Employee', 'Contractor']
    }

df = pd.DataFrame(mydict)

# Create column with total number of staff Status per Department
df['total_dept'] = df.groupby(['Department'])['Name'].transform('count')
print(df)
print('\n')


# Crate column with Status ratio per department
for k, v, in df.iterrows():
    df.loc[k, 'Status_Ratio'] = (df.groupby(['Department', 'Status']).count().xs(v['Status'], level=1)['total_dept'][v['Department']]/v['total_dept']) *100
print(df)
print('\n')

# Final Groupby with Status Ratio. Size NOT needed
print(df.groupby(['Department', 'Status', 'Status_Ratio']).size())

Desired Output:

Department  Status      Status_Ratio
Finance     Contractor  100.00
Marketing   Contractor  50.00
            Employee    50.00
Sales       Contractor  33.33 
            Employee    66.67

Upvotes: 2

Views: 4009

Answers (1)

Georgina Skibinski
Georgina Skibinski

Reputation: 13387

Try (with the original df):

df.groupby("Department")["Status"].value_counts(normalize=True).mul(100)

Outputs:

Department  Status
Finance     Contractor    100.000000
Marketing   Contractor     50.000000
            Employee       50.000000
Sales       Employee       66.666667
            Contractor     33.333333
Name: Status, dtype: float64

Upvotes: 5

Related Questions