Pandas iterate over group with SeriesGroupBy objects to Burst Data in MatPlotLib - python-3.x

Question

I am attempting to iterate through an index of a grouped-by dataframe (see comments in code).
import pandas as pd
import numpy as np
df = pd.DataFrame({'COL1':['A','A','B','B'], 'COL2': [1,1,2,2,], 'COL3': [2,3,4,6]})
#Here, I'm creating a copy of 'COL2' so that it's still a column after I assign it to
#the index.
#I'm guessing there's a better way to do this in the groupby line.
df['COL2_copy'] = df['COL2']
df = df.groupby(['COL2_copy'], as_index=True)
#This will actually be a more complex function (creating a chart in MatPlotLib based on the
#data frame)
#slice(group) per iteration through the index ('COL2')).
#I'll just keep it simple for now.
def pandfun(df):
#Here's the real issue:
df['COL4'] = np.trunc(np.round(df['COL3']*100,0))
return df['COL4']
pandfun(df)
TypeError: unsupported operand type(s) for *: 'SeriesGroupBy' and 'int'
The desired results are: 200 and 300 for the first group and 400 and 600 for the second group. So to summarize what I believe to be the main problem here is that I want to select individual groups of rows by index (i.e. 'COL2 == 1') and within each group, refer to individual rows for a calculation.
I am taking this approach because I'll actually be using this with a MatPlotLib function that I created and I want to "burst" the data into one chart for each group in the dataframe, where each chart refers to individual row data for a given group.

Answer 1

I did this instead:
1. Get a unique list of values from COL2 and create a copy of df:
ulist = pd.unique(df['COL2'].ravel())
df = df2
Iterate over that list where it matched in COL2:
for i in ulist:
df = df2.loc[df2['COL2']==i]
Within each iteration, apply the function.
Within the MatPlotLib function, enter the following code at the top:
df.reset_index()
This served to reset the selection of rows after each iteration.