Problems using retime within timetable

Dear all,

Using daily zero coupon yields data from 16-Aug-1972 until 13-Jul-2018 (see Gurkaynak et al 2006), I used

rmissing(retime(data,'frequency',@mean))

to get monthly and quarterly averages. The resulting monthly averages were fine. The same holds for the quarterly averages, however, with the exemption that matlab skipped the calculation of 2008Q1. Only when first calculating monthly averages, all quarters were calculated.

Can someone educate me for why taking quarterly averages directly from daily worked for all quarters but 2008Q1?

Thanks. Best, Gavin

Answer 1

[Akira Agata]

I think your data set is the following:

https://www.federalreserve.gov/pubs/feds/2006/200628/200628abs.html

Looking at the data, there is NaN at 2008-03-21 in the data set (feds200628.xls) and this causes your problem. In default settings, the nanflag of the function mean is set to 'includenan'. So, the mean value of 2008Q1 was not calculated.

To avoid this, please use

retime(data,'quarterly','mean')

or

retime(data,'quarterly',@(x) mean(x, 'omitnan'))

[Gavin Goy]

I see. You're right, both works perfectly fine. Thanks!

[SEETHA JAYAN]

1)Wheather both gives results of nanmean.(by omiting mean)?

2)what is x over second code

[Akira Agata]

Hi Seetha-san,

Yes, both gives the results of nanmean ( = mean with 'omitnan' option).

And 'x' in the second code is the input variable of anonymous function. I would recommend accessing the following documentation page for more details on 'anonymous function'.

https://jp.mathworks.com/help/matlab/matlab_prog/anonymous-functions.html