熊猫数据框分组汇总

Question

I have a Pandas dataframe with thousands of rows, and these cols:

Name    Job   Department   Salary    Date 

I want to return a new df with two cols:

Unique_Job     Avg_Salary

The code I use to accomplish this:

jobs = df.groupby(['Job'])
dict = {}
for a,b in jobs:
    dict.update({a: b['Salary'].mean()})
dfJobs = pd.DataFrame(dict.items(), columns=['Unique_Job', 'Avg Salary'])

However, I know there must be a better way. Ideas? Thanks.

Answer 1

Yes, use the aggregate method of the groupby object.

jobs = df.groupby('Job').aggregate({'Salary': 'mean'})

There's even the mean method as shortcut:

jobs = df.groupby('Job')['Salary'].mean()

See http://pandas.pydata.org/pandas-docs/stable/groupby.html for more info and lots of examples

Answer 2

As you already have the means, I guess you struggle with making the new dataframe from the series, you get as the output. You can use Series.to_frame() and DataFrame.reset_index() methods to make the dataframe with two columns and then you only rename the columns. Like this:

jobs = df.groupby('Job')['Salary'].mean()
jobs = jobs.to_frame().reset_index()
jobs.columns = ['Unique_Job', 'Avg_Salary']