0

I have a data frame of daily data like following:

  Date        bhpau_o bhpau_c_1d  bhpau_l  bbl_o bbl_c_1d  bbl_l    blt_o blt_c_1d
  2007-01-01 19.89669   19.93834 19.94399 18.585   18.585 18.600 18.28723 18.39005
  2007-01-02 20.17571   19.94399 20.32699 18.585   18.585 18.600 18.59225 18.28723
  2007-01-03 20.42895   20.32699 20.15213 18.510   18.600 18.080 18.62441 18.98699
  2007-01-04 19.45845   20.15213 19.15969 17.525   18.080 17.450 17.75146 18.19537
  2007-01-05 18.79020   19.15969 18.82920 17.445   17.450 17.275 17.26992 17.49870
  2007-01-08 18.67068   18.82920 18.63943 17.380   17.275 17.355 17.09140 17.05766
  2007-01-09 18.78240   18.63943 19.26600 17.405   17.355 17.540 17.50670 17.23673
  2007-01-10 18.80098   19.26600 18.65337 17.350   17.540 17.500 17.08860 17.22542
  2007-01-11 19.06616   18.65337 19.23807 17.830   17.500 18.200 17.58981 17.13693
  2007-01-12 19.62668   19.23807 19.66585 17.950   18.200 18.330 18.12275 18.08571
  2007-01-15 19.74896   19.66585 19.79600 17.950   18.200 18.330 18.26427 18.06403

What I wish to do is apply calculation on subsets of a dataset. Specifically, I wish to do the following calculation:

  1. Compute the absolute value of daily overnight return of bbl (bbl_l-bbl_o)

  2. Compute the sum of such absolute values(obtained in step 1) with a 10-day period

  3. Compute the weights of each overnight return

  4. Compute the overnight return of bhpau (bhpau_l-bhpau_o) then divided by the overnight return of bbl(non-absolute value obtained in step 1)

  5. Multiple the weights(obtained from step 3) by the value from step 4

  6. Sum up every 10 days' products(from step 5)

The code I used to calculate WPC is as shown below. However, they are only for calculating a pooled WPC.

##calculating WPC
WF1 = abs(data$bbl_l-data$bbl_o)/sum(abs(data$bbl_l-data$bbl_o)) 
CON1 = (data$bhpau_o-data$bhpau_c_1d)/(data$bbl_l-data$bbl_o)
WPC1 = sum(WF1*CON1)

How can I loop the function and get WPCs for every 10 days? I tried rollapply but it ended up with error message. Any help will be appreciated.

cactussss
  • 705
  • 2
  • 6
  • 10
  • Reproducible example please.You can use lapply to put the datasets in lists and use lapply to de the regression as well. – Jonas Tundo Aug 19 '13 at 11:27
  • I have edit the code with a simply example. Could you please explain a bit more about how to use lapply to loop the regression? Thank you. – cactussss Aug 22 '13 at 12:43
  • This is not reproducible. Provide at least a part of your data. See http://stackoverflow.com/questions/5963269/how-to-make-a-great-r-reproducible-example – Jonas Tundo Aug 22 '13 at 13:46
  • JT85 Question has been edited. Thank you for your help. – cactussss Aug 22 '13 at 14:13

0 Answers0