141207

Pandas


相関と共分散

>>> import pandas.io.data as web
>>> for ticker in ['AAPL', 'IBM', 'MSFT', 'GOOG']:
...     try:
...         all_data[ticker] = web.get_data_yahoo(ticker, '1/1/2005', '1/1/2010')
...         price = DataFrame({tic: data['Adj Close'] for tic, data in all_data.
iteritems()})
...         volume = DataFrame({tic: data['Volume'] for tic, data in all_data.it
eritems()})
...     except:
...         print "Cant find ", ticker
...
Cant find  GOOG
>>> returns = price.pct_change()
>>> returns.tail()
                AAPL       IBM      MSFT
Date
2009-12-24  0.034382  0.004402  0.002582
2009-12-28  0.012376  0.013315  0.005519
2009-12-29 -0.011876 -0.003410  0.006952
2009-12-30  0.012018  0.005424 -0.013808
2009-12-31 -0.004191 -0.012616 -0.015475
>>> returns.corr()
          AAPL       IBM      MSFT
AAPL  1.000000  0.497221  0.445868
IBM   0.497221  1.000000  0.559678
MSFT  0.445868  0.559678  1.000000
>>> returns.cov()
          AAPL       IBM      MSFT
AAPL  0.000716  0.000206  0.000236
IBM   0.000206  0.000240  0.000171
MSFT  0.000236  0.000171  0.000390
>>> returns.corrwith(volume)
AAPL   -0.010122
IBM     0.031356
MSFT   -0.056822
dtype: float64