Pandas Pivot Multi Index

pandas_pivot_multiindex.py

x = pd.DataFrame({"Date": np.repeat(pd.date_range("2015-01-01", "2015-01-04"), 2), 
                  "Class": [1, 2, 1, 2, 1, 2, 1, 2], 
                  "Value": np.random.random(size=8)})

p1 = pd.pivot_table(x, index="Date", columns="Class")
p2 = pd.pivot_table(x, index="Date", columns="Class", values="Value")

p1
#               Value          
# Class              1         2
# Date                          
# 2015-01-01  0.440874  0.692439
# 2015-01-02  0.263481  0.246643
# 2015-01-03  0.583451  0.764958
# 2015-01-04  0.344014  0.781684

# Where we have not Explicitly set the values="<some column name>" 
# there is a MultiIndex on the columns 
p1.columns
# MultiIndex(levels=[[u'Value'], [1, 2]],
#           labels=[[0, 0], [0, 1]],
#           names=[None, u'Class'])

p2
# Class              1         2
# Date                          
# 2015-01-01  0.440874  0.692439
# 2015-01-02  0.263481  0.246643
# 2015-01-03  0.583451  0.764958
# 2015-01-04  0.344014  0.781684

# Where we do Explicitly set the values="<some column name>" 
# we have a single level index
p2.columns
#Int64Index([1, 2], dtype='int64', name=u'Class')