1

Ok, enough is enough. I need help with this iteritems() and append() procedeure...

Here we have some Time Series Price Data for barrels of Beer, and Whiskey...

    Beer    Whiskey
Date        
1978-12-29  22.60   86.50
1979-01-02  22.68   86.52
1979-01-03  21.86   87.41
1979-01-04  22.32   87.54
1979-01-05  22.55   87.49
1979-01-08  22.31   87.21
1979-01-09  22.40   87.61
1979-01-10  22.07   87.64
1979-01-11  22.07   88.12
1979-01-12  21.76   88.04

What I am trying to do is create rolling 5 day return values from this data. I have been using the iteritems() function and I am getting the right numbers. The first part that I don't understand is why this function repeats the output as many times as there are columns in the DataFrame (minus the index). This is the code and output...

for value in test.iteritems():
    print(((test - test.shift(5))*100)/test.shift(5))

OUTPUT

               Beer        Whiskey
Date                          
1978-12-29       NaN       NaN
1979-01-02       NaN       NaN
1979-01-03       NaN       NaN
1979-01-04       NaN       NaN
1979-01-05       NaN       NaN
1979-01-08 -1.283186  0.820809
1979-01-09 -1.234568  1.259824
1979-01-10  0.960659  0.263128
1979-01-11 -1.120072  0.662554
1979-01-12 -3.503326  0.628643
                Beer        Whiskey
Date                          
1978-12-29       NaN       NaN
1979-01-02       NaN       NaN
1979-01-03       NaN       NaN
1979-01-04       NaN       NaN
1979-01-05       NaN       NaN
1979-01-08 -1.283186  0.820809
1979-01-09 -1.234568  1.259824
1979-01-10  0.960659  0.263128
1979-01-11 -1.120072  0.662554
1979-01-12 -3.503326  0.628643

Any ideas why this exact output is repeated?

NEXT, I create a new DataFrame and I ask (very nicely!) to append this output into the dataframe. Here is the code...

for value in test.iteritems():
    df.append(((test - test.shift(5))*100)/test.shift(5))

This is the error I receive...

TypeError                                 Traceback (most recent call last)
<ipython-input-133-006bdc416056> in <module>()
      1 for value in test.iteritems():
----> 2     df.append(((test - test.shift(5))*100)/test.shift(5))

TypeError: append() missing 1 required positional argument: 'other'

My research says that this 'other' TypeError occurs when there is a reference missing in the code. I have tried different combinations of "key, value" with no avail. Further, the print function seems to not have any issues. Please let me know if you have any ideas. Thanks in advance

Ludlow_Luddite
  • 63
  • 8

1 Answers1

1

pandas.iteritems iterates over pairs of the form name, column (series to be more precise), you can check that by looking at this example

for value in test.iteritems():
    print(value[0])

This outputs

Beer
Whiskey

That's why you see multiple outputs of the same frame. A simple solution to your problem is

returns = 100 * test.diff(5) / test.shift(5)
print(returns)
                Beer   Whiskey
Date                          
1978-12-29       NaN       NaN
1979-01-02       NaN       NaN
1979-01-03       NaN       NaN
1979-01-04       NaN       NaN
1979-01-05       NaN       NaN
1979-01-08 -1.283186  0.820809
1979-01-09 -1.234568  1.259824
1979-01-10  0.960659  0.263128
1979-01-11 -1.120072  0.662554
1979-01-12 -3.503326  0.628643
caverac
  • 1,505
  • 2
  • 12
  • 17
  • Thanks caverac, I appreciate the reply. I have created a variable 'returns' as you suggested. Nonetheless, I continue to get the same dual output. I am not quite sure what you mean by, "pairs of the form name, column (series to be more precise)." – Ludlow_Luddite Dec 13 '18 at 11:13
  • @Ludlow_Luddite You don't need a loop. With the single statement `returns = ...` you calculate the returns for all the colums – caverac Dec 13 '18 at 11:15
  • That's awesome. It is too easy to make things more complicated than they need be. * df = returns * I love the diff() function too. Thanks! – Ludlow_Luddite Dec 14 '18 at 01:21
  • @Ludlow_Luddite Happy to help – caverac Dec 14 '18 at 01:22