Pandas Build empty SparseDataFrame by columns very loog compared to by index.

Code Sample, a copy-pastable example if possible

I want to create a sparse matrix with a 4 level multiindex and about 340 000 x 340 000 cells. I is not possible to build it in dense and to sparse it. So I tried to build it directly in SparseDataFrame.

n = len(index)
print(n)
>>> 338275
%timeit df0 = pd.SparseDataFrame(index=index)
1000 loops, best of 3: 419 µs per loop

But if I tried to construct:

df1 = pd.SparseDataFrame(columns=index)

or

df2 = pd.SparseDataFrame(index=index, columns=index)

An all night wasn't enough to build the empty SparseDataFrame. I don't understand how to build this empty SparseDataFrame in a quite reseanoble time (less than <20 minutes with 8 GoRam).

Output of pd.show_versions()

commit: None
python: 3.6.0.final.0
python-bits: 64
OS: Darwin
OS-release: 15.6.0
machine: x86_64
processor: i386
byteorder: little
LC_ALL: None
LANG: fr_FR.UTF-8
LOCALE: fr_FR.UTF-8

pandas: 0.19.2
nose: 1.3.7
pip: 9.0.1
setuptools: 27.2.0
Cython: 0.25.2
numpy: 1.11.3
scipy: 0.19.0
statsmodels: 0.6.1
xarray: None
IPython: 5.1.0
sphinx: 1.5.1
patsy: 0.4.1
dateutil: 2.6.0
pytz: 2016.10
blosc: None
bottleneck: 1.2.0
tables: 3.3.0
numexpr: 2.6.1
matplotlib: 2.0.0
openpyxl: 2.4.1
xlrd: 1.0.0
xlwt: 1.2.0
xlsxwriter: 0.9.6
lxml: 3.7.2
bs4: 4.5.3
html5lib: None
httplib2: None
apiclient: None
sqlalchemy: 1.1.5
pymysql: None
psycopg2: None
jinja2: 2.9.4
boto: 2.45.0

Comment From: jreback

can can try this (avail in 0.20.0 releasing shortly): http://pandas-docs.github.io/pandas-docs-travis/sparse.html#sparsedataframe.

though I suspect that may not work. pandas sparse is row based, NOT column based, so your size would blow up memory.

Comment From: cfrancois7

Thank you. Yes, I've already tried by importing a COO sparse matrix without success. Indeed, it makes sense my approach didn't work because pandas sparse is row based. I'm changing my approach by SpareSeries.