I have an xts object with 12 variables over time in 15 minute intervals.
> summary(wideRawXTSscaled)
Index DO0182U09A3 DO0182U09B3 DO0182U09C3 DO0182U21A1 DO0182U21A2
Min. :2017-01-20 16:30:00 Min. :-1.09338 Min. :-1.0666 Min. :-0.9700 Min. :-1.2687 Min. :-1.00676
1st Qu.:2017-01-24 04:22:30 1st Qu.:-0.60133 1st Qu.:-0.6675 1st Qu.:-0.6009 1st Qu.:-0.4522 1st Qu.:-0.48525
Median :2017-01-27 16:15:00 Median :-0.38317 Median :-0.2742 Median :-0.1761 Median :-0.2127 Median :-0.27482
Mean :2017-01-27 16:15:00 Mean : 0.00000 Mean : 0.0000 Mean : 0.0000 Mean : 0.0000 Mean : 0.00000
3rd Qu.:2017-01-31 04:07:30 3rd Qu.: 0.08221 3rd Qu.: 0.2922 3rd Qu.: 0.1125 3rd Qu.: 0.1248 3rd Qu.: 0.05455
Max. :2017-02-03 16:00:00 Max. : 3.33508 Max. : 9.2143 Max. : 5.8473 Max. :18.4909 Max. :12.21382
DO0182U21A3 DO0182U21B1 DO0182U21B2 DO0182U21B3 DO0182U21C1 DO0182U21C2 DO0182U21C3
Min. :-1.09339 Min. :-1.0268 Min. :-0.9797 Min. :-1.0853 Min. :-1.3556 Min. :-1.15469 Min. :-1.2063
1st Qu.:-0.33919 1st Qu.:-0.6020 1st Qu.:-0.5597 1st Qu.:-0.6692 1st Qu.:-0.5600 1st Qu.:-0.37291 1st Qu.:-0.3460
Median :-0.21082 Median :-0.3389 Median :-0.3466 Median :-0.3828 Median :-0.2138 Median :-0.16183 Median :-0.1635
Mean : 0.00000 Mean : 0.0000 Mean : 0.0000 Mean : 0.0000 Mean : 0.0000 Mean : 0.00000 Mean : 0.0000
3rd Qu.:-0.01826 3rd Qu.: 0.2105 3rd Qu.: 0.2486 3rd Qu.: 0.5624 3rd Qu.: 0.1992 3rd Qu.: 0.08052 3rd Qu.: 0.1363
Max. :12.69083 Max. : 7.7314 Max. : 9.2900 Max. : 7.3540 Max. :13.7427 Max. :13.76166 Max. :15.8086
I wish to compute the correlations between each of the variables for each interval of time. Since i have 12 variables I expect to have a 12 x 12 matrix for each of the 15 min data points in my xts object.
For the correlation computation I am using the following code:
wideRawXTSscaledCorr <- rollapplyr(wideRawXTSscaled, 10, cor, by.column = FALSE)
"10" in the code above uses 10 time series values to calculate the correlation matrix therefore I will have 9 NA values at the start of my wideRawXTSscaledCorr with the correlation values returned in the 10th.
> wideRawXTSscaledCorr[1:10,1:5]
[,1] [,2] [,3] [,4] [,5]
2017-01-20 16:30:00 NA NA NA NA NA
2017-01-20 16:45:00 NA NA NA NA NA
2017-01-20 17:00:00 NA NA NA NA NA
2017-01-20 17:15:00 NA NA NA NA NA
2017-01-20 17:30:00 NA NA NA NA NA
2017-01-20 17:45:00 NA NA NA NA NA
2017-01-20 18:00:00 NA NA NA NA NA
2017-01-20 18:15:00 NA NA NA NA NA
2017-01-20 18:30:00 NA NA NA NA NA
2017-01-20 18:45:00 1 0.1590656 0.2427391 0.1987761 -0.1026246
When I change the value of the sliding window to any values <10 I get repetitions of the following error:
> wideRawXTSscaledCorr <- rollapplyr(wideRawXTSscaled, 7, cor, by.column = FALSE)
Warning messages:
1: In FUN(.subset_xts(data, (i - width + 1):i), ...) :
the standard deviation is zero
2: In FUN(.subset_xts(data, (i - width + 1):i), ...) :
the standard deviation is zero
3: In FUN(.subset_xts(data, (i - width + 1):i), ...) :
the standard deviation is zero
4: In FUN(.subset_xts(data, (i - width + 1):i), ...) :
the standard deviation is zero
5: In FUN(.subset_xts(data, (i - width + 1):i), ...) :
the standard deviation is zero
Is this the only test I can use to see if these errors are solely a symptom of my data or could it be due to some coding error I have made? Is there some other way I can go deeper into the code to see which values are causing these errors?
