[ENH] Fama/French

[0x0L]

• Added get_datasets_famafrench (requires bs4) : list all available datasets from FF
• Complete rewrite of get_data_famafrench. Can grab and parse all (so called) csv files from FF website. Also uses metadata in the files to generate documentation (see DESCR attribute)
• Moved tests for FF to their own file (One file per datareader #59)

API change : indexes are now pandas.PeriodIndex for annual and monthly data, and pandas.DatetimeIndex otherwise.

Should fix #55 #20 #21 see also pandas-dev/pandas#8842

[0x0L]

This looks terrible

[davidastephens]

I can't think of any other way to do that unless we import dateutil.

[0x0L]

any substitute for findAll which does not involve any import ?

[davidastephens]

We already have a dependency on lxml, so maybe use that?

[davidastephens]

Check this gist out:
https://gist.github.com/yejianye/5895954

Also, we could pull the Options._parse_url method out into a common function, as it does the importing and error checking for lxml.

[0x0L]

@davidastephens done

[davidastephens]

Thanks. We will need to rebase/squash after #69 is merged, but other than that, looks good to me.

[davidastephens]

#69 merged, can you rebase?

[0x0L]

Done

I also added a test which fails on python 2.6.
It seems like it's a bug in pandas 0.13.1: docstring claims period inference should work but it doesn't

>>> df[0].to_period()
Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
  File "/opt/python/2.6.9/lib/python2.6/site-packages/pandas/core/frame.py", line 4258, in to_period
    freq = self.index.freqstr or self.index.inferred_freq
  File "/opt/python/2.6.9/lib/python2.6/site-packages/pandas/tseries/index.py", line 1393, in freqstr
    return self.offset.freqstr
AttributeError: 'NoneType' object has no attribute 'freqstr'
>>> df[0].to_period('M')
         Mkt-RF   SMB    HML    RF
1926-07    2.96 -2.30  -2.87  0.22
1926-08    2.64 -1.40   4.19  0.25
1926-09    0.36 -1.32   0.01  0.23
1926-10   -3.24  0.04   0.51 ...
...
>>> print df[0].to_period.__doc__

        Convert DataFrame from DatetimeIndex to PeriodIndex with desired
        frequency (inferred from index if not passed)

        Parameters
        ----------
        freq : string, default
        axis : {0, 1}, default 0
            The axis to convert (the index by default)
        copy : boolean, default True
            If False then underlying input data is not copied

        Returns
        -------
        ts : TimeSeries with PeriodIndex

[davidastephens]

@jreback @jorisvandenbossche Thoughts?

Need to keep supporting pandas 13.1 or move that one to an 'allowed failure'?

[0x0L]

@davidastephens i wrote a workaround for 13.1
(tests pass for famafrench, failed on Fred on HTTP 503 error)

[davidastephens]

Thanks!