python, pandas, dataframe, rows to columns

Question

I've got a dataframe I pulled from a poorly organized SQL table. That table has unique rows for every channel I can extract that info to a python dataframe, and intend to do further processing, but for now just want to get it to a more usable format

sample input:

C = pd.DataFrame()
A = np.array([datetime.datetime(2016,8,8,0,0,1,1000),45,'foo1',1])
B = pd.DataFrame(A.reshape(1,4),columns = ['date','chNum','chNam','value'])
C = C.append(B)
A = np.array([datetime.datetime(2016,8,8,0,0,1,1000),46,'foo2',12.3])
B = pd.DataFrame(A.reshape(1,4),columns = ['date','chNum','chNam','value'])
C = C.append(B)
A = np.array([datetime.datetime(2016,8,8,0,0,2,1000),45,'foo1',10])
B = pd.DataFrame(A.reshape(1,4),columns = ['date','chNum','chNam','value'])
C = C.append(B)
A = np.array([datetime.datetime(2016,8,8,0,0,2,1000),46,'foo2',11.3])
B = pd.DataFrame(A.reshape(1,4),columns = ['date','chNum','chNam','value'])
C = C.append(B)

Produces

                             date chNum chNam value
0  2016-08-08 00:00:01.001000    45  foo1     1
0  2016-08-08 00:00:01.001000    46  foo2  12.3
0  2016-08-08 00:00:02.001000    45  foo1    10
0  2016-08-08 00:00:02.001000    46  foo2  11.3

I want

                                 date foo1     foo2  
2016-08-08 00:00:01.001000           1     12.3
2016-08-08 00:00:02.001000           10   113

I have a solution: make a list of unique dates, for each date loop through the dataframe and pull off each channel, making a new row. kind of tedious (error prone)! to program, so I was wondering if there's a better way to utilize Pandas tools

Answer 1

Use set_index then unstack to pivot

C.set_index(['date', 'chNum', 'chNam'])['value'].unstack(['chNam', 'chNum'])

---

To get exactly what you asked for

C.set_index(['date', 'chNam'])['value'].unstack().rename_axis(None, 1)