BUG: Google finance is returning object dtype data (rather than float)

[jreback]

No description provided.

[cpcloud]

do u want to put this under cleanup data.py? i can take it on

[cpcloud]

or rather should i mention it in the pr?

[jreback]

u can move 2 your cleanup of data
but can leave most/all of that till 0.12

[jreback]

I think this just needs a conversion:

In [28]: x = web.get_data_google(['IBM','INTC'],'1/1/2010','1/1/2013')
x

In [29]: x
Out[29]: 
<class 'pandas.core.panel.Panel'>
Dimensions: 5 (items) x 761 (major_axis) x 2 (minor_axis)
Items axis: Open to Volume
Major_axis axis: 2010-01-01 00:00:00 to 2012-12-31 00:00:00
Minor_axis axis: IBM to INTC

In [30]: x.loc[:,:,'IBM']
Out[30]: 
<class 'pandas.core.frame.DataFrame'>
DatetimeIndex: 761 entries, 2010-01-01 00:00:00 to 2012-12-31 00:00:00
Data columns (total 5 columns):
Open      761  non-null values
High      761  non-null values
Low       761  non-null values
Close     761  non-null values
Volume    761  non-null values
dtypes: object(5)

In [31]: x.loc[:,:,'IBM'].head()
Out[31]: 
              Open    High     Low   Close   Volume
Date                                              
2010-01-01       -       -       -   130.9        0
2010-01-04  131.18  132.97  130.85  132.45  6155846
2010-01-05  131.68  131.85  130.10  130.85  6842471
2010-01-06  130.68  131.49  129.81     130  5605290
2010-01-07  129.87  130.25  128.91  129.55  5840569

In [32]: x.loc[:,:,'IBM'].dtypes
Out[32]: 
Open      object
High      object
Low       object
Close     object
Volume    object
dtype: object

In [34]: x.loc[:,:,'IBM'].convert_objects(convert_numeric='force')
Out[34]: 
<class 'pandas.core.frame.DataFrame'>
DatetimeIndex: 761 entries, 2010-01-01 00:00:00 to 2012-12-31 00:00:00
Data columns (total 5 columns):
Open      754  non-null values
High      754  non-null values
Low       754  non-null values
Close     761  non-null values
Volume    761  non-null values
dtypes: float64(4), int64(1)

In [35]: x.loc[:,:,'IBM'].convert_objects(convert_numeric='force').head()
Out[35]: 
              Open    High     Low   Close   Volume
Date                                              
2010-01-01     NaN     NaN     NaN  130.90        0
2010-01-04  131.18  132.97  130.85  132.45  6155846
2010-01-05  131.68  131.85  130.10  130.85  6842471
2010-01-06  130.68  131.49  129.81  130.00  5605290
2010-01-07  129.87  130.25  128.91  129.55  5840569

[davidastephens]

This looks to have been fixed. get_data_google returns floats instead of objects.

x = web.get_data_google(['IBM'],'1/1/2010','1/1/2014')

x.loc[:,:,'IBM'].head()
Out[19]: 
              Open    High     Low   Close   Volume
Date                                              
2010-01-04  131.18  132.97  130.85  132.45  6155846
2010-01-05  131.68  131.85  130.10  130.85  6842471
2010-01-06  130.68  131.49  129.81  130.00  5605290
2010-01-07  129.87  130.25  128.91  129.55  5840569
2010-01-08  129.07  130.92  129.05  130.85  4197105

x.dtypes
Out[20]: 
Open      float64
High      float64
Low       float64
Close     float64
Volume    float64
dtype: object

[jreback]

@dstephens99 can you put up a validation test so we can close?

[davidastephens]

Sure. Do you want a mention in whats new as well?

[jreback]

sure you can mention. Also if you can figure out when this was fixed (e.g. use git blame), you can say, fixed in version 0.14.1...or whatever (if not ok too). But goes in 0.15.2