CODEWITHSUNDEEP
pythonpandas
Steve Lorimer
Steve Lorimer

Reputation: 28659

Transform pandas panel into dataframe

I have a pandas.Panel, and I want to create a pandas.DataFrame with the column headers coming from one column, the data from another column, and the number of rows is the number of items in the panel.

If diagrams will help describe what I'm looking for...

My panel looks somewhat like the following:

      +---+---------+------------+------+
    +---+---------+------------+------+ |
  +---+---------+------------+------+ |-+
+---+---------+------------+------+ |-+ |
|   | context | iterations | time |-+ |-+
+---+---------+------------+------+ |-+ |
| 0 | foo     |          1 |   21 |-+ |-+
+---+---------+------------+------+ |-+ |
| 1 | bar     |          2 |   37 |-+ |-+
+---+---------+------------+------+ |-+
| 2 | baz     |          1 |   53 |-+
+---+---------+------------+------+

I would like to transform the panel into a dataframe:

The result would look something like this:

+---+-----+-----+-----+
|   | foo | bar | baz |
+---+-----+-----+-----+
| 0 |  21 |  37 |  53 |
+---+-----+-----+-----+
| 1 |  36 |  42 |  76 |
+---+-----+-----+-----+
| 2 |  24 |  56 |  83 |
+---+-----+-----+-----+
| 3 |  17 |  32 |  45 |
+---+-----+-----+-----+

Upvotes: 3

Views: 741

Answers (1)

piRSquared
piRSquared

Reputation: 294218

Option 1
pd.concat 

pd.concat({i: d.set_index('context').time for i, d in pn.iteritems()}).unstack()

context  foo  bar  baz
0         21   37   53
1         36   42   76
2         24   56   83
3         17   32   45

Option 2
pd.DataFrame 

pd.DataFrame([d.set_index('context').time for i, d in pn.iteritems()], pn.items)

context  foo  bar  baz
0         21   37   53
1         36   42   76
2         24   56   83
3         17   32   45

Upvotes: 1

Related Questions