[R] Missing data?

Kevin Burton rkevinburton at charter.net
Wed Nov 23 00:50:45 CET 2011


Void of any other suggestions this approach makes sense but for my case I
think I need to use zoo objects rather than xts. If I sequence the data
generally I don't know if there will be 365 days in the year or 366. So I
have to sequence the dates as:

seq(from=as.Date("2011-01-01"), to=as.Date("2011-12-31"), by="day")

If I use this sequence with xts I get:

> ds <- xts(NA, seq(from=as.Date("2011-01-01"), to=as.Date("2011-12-31"),
by="day"))
Error in xts(NA, seq(from = as.Date("2011-01-01"), to =
as.Date("2011-12-31"),  : 
  NROW(x) must match length(order.by)

If I leave the 'data' empty I don't get the error but if I try to assign an
individual item (fill as appropriate)

> ds <- xts(, seq(from=as.Date("2011-01-01"), to=as.Date("2011-12-31"),
by="day"))
> ds["2011-12-24"] <- 10
> ds
Error in structure(coredata(x), names = x.attr$dimnames[[1]]) : 
  'names' attribute [365] must be the same length as the vector [358]

So now I need to remember that I have not filled in all of the data. Also
simple dereferencing gives:

> ds[1]
Error in `[.xts`(ds, 1) : subscript out of bounds

With zoo I am able to create a time-series where all of the data is
initially NA:

> ds <- zoo(NA, seq(from=as.Date("2011-01-01"), to=as.Date("2011-12-31"),
by="day"))

So I can fill the data as appropriate and the remaining slots will have NA.
I may be new with xts but I cannot see a way of creating a useable 'blank'
time-series.

Also with xts it seems like the frequency is ignored.

> ds <- xts(1:365, seq(from=as.Date("2011-01-01"), to=as.Date("2011-12-31"),
by="day"), frequency=52)
> frequency(ds)
[1] 1

Whereas zoo remembers the frequency setting

> ds <- zoo(1:365, seq(from=as.Date("2011-01-01"), to=as.Date("2011-12-31"),
by="day"), frequency=52)
> frequency(ds)
[1] 52

But since the ultimate goal is to get the time-series in a 'ts' format (as
many functions require 'ts') it seems like even zoo has problems:

> as.ts(ds)

Time Series:
Start = c(14975, 1) 
End = c(15339, 1) 
Frequency = 52 
    [1]   1  NA  NA  NA  NA  NA  NA  NA  NA  NA  NA  NA  NA  NA  NA  NA  NA
NA  NA  NA  NA  NA  NA  NA  NA  NA  NA  NA  NA  NA  NA  NA  NA  NA  NA  NA
NA  NA  NA  NA  NA
   [42]  NA  NA  NA  NA  NA  NA  NA  NA  NA  NA  NA   2  NA  NA  NA  NA  NA
NA  NA  NA  NA  NA  NA  NA  NA  NA  NA  NA  NA  NA  NA  NA  NA  NA  NA  NA
NA  NA  NA  NA  NA
   [83]  NA  NA  NA  NA  NA  NA  NA  NA  NA  NA  NA  NA  NA  NA  NA  NA  NA
NA  NA  NA  NA  NA   3  NA  NA  NA  NA  NA  NA  NA  NA  NA  NA  NA  NA  NA
NA  NA  NA  NA  NA
  [124]  NA  NA  NA  NA  NA  NA  NA  NA  NA  NA  NA  NA  NA  NA  NA  NA  NA
NA  NA  NA  NA  NA  NA  NA  NA  NA  NA  NA  NA  NA  NA  NA  NA   4  NA  NA
NA  NA  NA  NA  NA
  [165]  NA  NA  NA  NA  NA  NA  NA  NA  NA  NA  NA  NA  NA  NA  NA  NA  NA
NA  NA  NA  NA  NA  NA  NA  NA  NA  NA  NA  NA  NA  NA  NA  NA  NA  NA  NA
NA  NA  NA  NA  NA
  [206] . . . . . .
 So the conversion from zoo to ts maintained the frequency but I am not sure
where it decided on the start and end values. Also the conversion seemed to
changed the data also. Notice that every period (52 entries) the original
data is maintained. In other words if ds is the original zoo time series
then ds[1] is 1 and ds[2] is 2 etc. The converted time-series keeps ds[1]
but inserts 51 NA's then adds ds[2] etc till the end of the series.  That is
not what the initial data was. The conversion is inserting data of its own.

The conversion to ts from xts seems better behaved:

ds <- xts(1:365, seq(from=as.Date("2011-01-01"), to=as.Date("2011-12-31"),
by="day"), frequency=52)
> as.ts(ds)
Time Series:
Start = 1 
End = 365 
Frequency = 1 
  [1]   1   2   3   4   5   6   7   8   9  10  11  12  13  14  15  16  17
18  19  20  21  22  23  24  25  26  27  28  29  30  31  32  33  34  35  36
37  38  39  40  41  42
 [43]  43  44  45  46  47  48  49  50  51  52  53  54  55  56  57  58  59
60  61  62  63  64  65  66  67  68  69  70  71  72  73  74  75  76  77  78
79  80  81  82  83  84
 [85]  85  86  87  88  89  90  91  92  93  94  95  96  97  98  99 100 101
102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120
121 122 123 124 125 126
[127] 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143
144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162
163 164 165 166 167 168
[169] 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185
186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204
205 206 207 208 209 210
[211] 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227
228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246
247 248 249 250 251 252
[253] 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269
270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288
289 290 291 292 293 294
[295] 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311
312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330
331 332 333 334 335 336
[337] 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353
354 355 356 357 358 359 360 361 362 363 364 365

But alas the frequency is ignored.

So this is what I have found out using these two packages.  If I want to
create a 'blank' data set it seems like zoo is 'better' since I can create a
time-series initialized with NA irrespective of the length of the series.
However I must be unfamiliar with the conversion because zoo doesn't convert
to a regular 'ts' very well.  But  zoo remembers the frequency setting
whereas xts just ignores it.

It seems like there is still considerable work to solve the original
problem. If I create a time series and fill in the values that are
appropriate I still could have NA in the series it seems to.weekly has a
problem with NA in the time series:
> ds <- xts(rep(NA,365), seq(from=as.Date("2011-01-01"),
to=as.Date("2011-12-31"), by="day"), frequency=52)
> to.weekly(ds, sum)
Error in if (drop.time) x <- .drop.time(x) : 
  argument is not interpretable as logical
In addition: Warning message:
In to.period(x, "weeks", name = name, ...) :
  missing values removed from data


-----Original Message-----
From: R. Michael Weylandt <michael.weylandt at gmail.com>
[mailto:michael.weylandt at gmail.com] 
Sent: Tuesday, November 22, 2011 3:10 PM
To: Kevin Burton
Cc: <r-help at r-project.org>
Subject: Re: [R] Missing data?

Couldn't you use seq.Date() to set up the time index and then just fill as
appropriate?

Alternatively, to.weekly if you are starting with a daily series. 

Michael

On Nov 22, 2011, at 4:00 PM, "Kevin Burton" <rkevinburton at charter.net>
wrote:

> I was wondering what the best approach is for missing data in a time
series.
> I give an example using xts but I would like to know what seems to be 
> the "best" method. Say I have
> 
> 
> 
> library(xts)
> 
> xts.ts <- xts(1:4,as.Date(c("1970-01-01", "1970-1-3", "1980-10-10", 
> "2007-8-19")), frequency=52)
> 
> 
> 
> I would like to turn this into a time series (still could be xts, or 
> converted to ts) that has values for every week starting with the week 
> that includes the start date and ending with the week that includes the
end date.
> If there is data for the week then use it otherwise set it to NA or 0.
> Remember some years have 52, 53, or rarely 54 full or partial weeks. 
> What to do with the partials at the beginning and ending of the year? 
> This seems to be a fairly common problem and doing it myself is very 
> cumbersome. Does a solution to this kind of problem exist? Once the 
> approach to a weekly period is found I am sure that adjustment to 
> daily, monthly, or quarterly would be relatively straightforward.
> 
> 
> 
> Thank you.
> 
> 
> 
> Kevin
> 
> 
> 
> 
>    [[alternative HTML version deleted]]
> 
> ______________________________________________
> R-help at r-project.org mailing list
> https://stat.ethz.ch/mailman/listinfo/r-help
> PLEASE do read the posting guide 
> http://www.R-project.org/posting-guide.html
> and provide commented, minimal, self-contained, reproducible code.



More information about the R-help mailing list