在pandas DataFrame中为列接收NaN

Question

This is a piece of code (exercise) from the O'Reilly book Python for Data Analysis .

from pandas import Series, DataFrame
import pandas.io.data as web

all_data = {}
for ticker in ['AAPL', 'IBM', 'MSFT', 'GOOG']:
    all_data[ticker] = web.get_data_yahoo(ticker)

price = DataFrame({k: v['Adj Close'] for k,v in all_data.items()})

The strange this is that when I look at the resulting DataFrame, the contents for Google are always NaN :

I know that the code is not what you'd call optimal, but these are book exercises and I'm trying to learn from them by experimenting.

If I take only the data relating to Google and make a DataFrame out of that, the actual figures appear:

DataFrame(all_data['GOOG']['Adj Close']).head()

But when I try to do the same thing for all ticker symbols, it goes wrong again:

DataFrame([all_data['GOOG']['Adj Close'],
         all_data['AAPL']['Adj Close'],
         all_data['IBM']['Adj Close'],
         all_data['MSFT']['Adj Close']],
         index=['GOOG', 'AAPL', 'IBM', 'MSFT']).T.head()

Any insight as to what might be causing this would be greatly appreciated!

Version info:

• Python 3.4.2
• pandas (0.16.2)
• numpy (1.9.2)

Answer 1

Google now has two classes of publicly traded stock, the class C ("GOOG") was issued in 2014, the original A shares trade under "GOOGL". Article here with some more info.

So to have the complete history for all 4, just change the ticker. This also is a pretty good example of what it means for data to "missing". If you wanted to filter to common dates for those original 4 tickers you could do price = price.dropna()

Answer 2

you are not looking at the full data. Look at the dates in your two rearrangements.

>>> price.GOOG.isnull().sum()
1064

try tail()

>>> price.GOOG.head()
Date
2010-01-04   NaN
2010-01-05   NaN
2010-01-06   NaN
2010-01-07   NaN
2010-01-08   NaN

>>> price.GOOG.tail()
Date
2015-08-24    589.609985
2015-08-25    582.059998
2015-08-26    628.619995
2015-08-27    637.609985
2015-08-28    630.380005

I suspect the underlying reason is a RIC change on the part of google. They have changed their share structure several times to keep control of voting rights etc. So the stock price is not defined for that stock identifier before a certain date.

It might help to use an IDE like Spyder - you can view the full data frame in a matlab like way, which stops this kind of thing happening.