The following code snippet
df = pandas.DataFrame({})
for run in range(3):
df[str(run)] = pandas.Series([1,2])
produces an empty dataframe (instead of three columns of [1,2], indexed by 0, 1, 2). Why, and how do I fix this?
More precisely, this is what is output by pandas version 0.11 (run in ipython with Python 2.7):
In [17]: df
Out[17]:
Empty DataFrame
Columns: [0, 1, 2]
Index: []
Works fine in 0.13
In [1]: df = pandas.DataFrame({})
In [2]: for run in range(3):
...: df[str(run)] = pandas.Series([1,2])
...:
In [3]: df
Out[3]:
0 1 2
0 1 1 1
1 2 2 2
[2 rows x 3 columns]
Much more efficient to do this, however
In [8]: df = pd.concat([ pandas.Series([1,2]) for i in range(3) ], axis=1)
In [9]: df
Out[9]:
0 1 2
0 1 1 1
1 2 2 2
[2 rows x 3 columns]
I can't reproduce. Try again.
In [8]: df = DataFrame({})
In [9]: for run in range(3):
df[str(run)] = Series([1,2])
...:
In [10]: df
Out[10]:
0 1 2
0 1 1 1
1 2 2 2
[2 rows x 3 columns]
If it fails, what version of pandas? (Though I'm almost certain this is fine in any version.)
