Detect and replace omitted data

I am analyzing the spatial distribution of fish in a stream. The stream is
divided into equally sized units, and the number of fish in each unit is
counted. My problem is that my dataset is missing rows where the count in a
unit equals zero. I need to create zero data for the missing units.

For example:
day<-(c(rep(4,8),rep(6,8)))
unit<-c(seq(1,8,1),seq(2,16,2))
value<-floor(rnorm(16,25,10))
x<-cbind(day,unit,value)

x
? ? ?day unit value
?[1,] ? 4 ? ?1 ? ?19
?[2,] ? 4 ? ?2 ? ?15
?[3,] ? 4 ? ?3 ? ?16
?[4,] ? 4 ? ?4 ? ?20
?[5,] ? 4 ? ?5 ? ?17
?[6,] ? 4 ? ?6 ? ?15
?[7,] ? 4 ? ?7 ? ?14
?[8,] ? 4 ? ?8 ? ?29
?[9,] ? 6 ? ?2 ? ?18
[10,] ? 6 ? ?4 ? ?22
[11,] ? 6 ? ?6 ? ?27
[12,] ? 6 ? ?8 ? ?16
[13,] ? 6 ? 10 ? ?45
[14,] ? 6 ? 12 ? ?36
[15,] ? 6 ? 14 ? ?34
[16,] ? 6 ? 16 ? ?13

Lets say the stream has 16 units. For each day, I want to fill in rows for
any missing units (e.g., units 9-16 for day 4, the odd numbered units on day
6) with values of zero.

Does anyone know a relatively concise way to do this?
Thank you.

? ? ? ?[[alternative HTML version deleted]]

Hi Jonny,

On Tue, Oct 18, 2011 at 1:02 PM, Jonny Armstrong
<jonny5armstrong at gmail.com> wrote:

I am analyzing the spatial distribution of fish in a stream. The  
stream is
divided into equally sized units, and the number of fish in each  
unit is
counted. My problem is that my dataset is missing rows where the  
count in a
unit equals zero. I need to create zero data for the missing units.

For example:
day<-(c(rep(4,8),rep(6,8)))
unit<-c(seq(1,8,1),seq(2,16,2))
value<-floor(rnorm(16,25,10))
x<-cbind(day,unit,value)

Thanks for the actual reproducible example.

x
     day unit value
 [1,]   4    1    19
 [2,]   4    2    15
 [3,]   4    3    16
 [4,]   4    4    20
 [5,]   4    5    17
 [6,]   4    6    15
 [7,]   4    7    14
 [8,]   4    8    29
 [9,]   6    2    18
[10,]   6    4    22
[11,]   6    6    27
[12,]   6    8    16
[13,]   6   10    45
[14,]   6   12    36
[15,]   6   14    34
[16,]   6   16    13

Lets say the stream has 16 units. For each day, I want to fill in  
rows for
any missing units (e.g., units 9-16 for day 4, the odd numbered  
units on day
6) with values of zero.

On Oct 18, 2011, at 2:24 PM, Sarah Goslee wrote:

Hi Jonny,

On Tue, Oct 18, 2011 at 1:02 PM, Jonny Armstrong
<jonny5armstrong at gmail.com> wrote:

I am analyzing the spatial distribution of fish in a stream. The stream
is
divided into equally sized units, and the number of fish in each unit is
counted. My problem is that my dataset is missing rows where the count in
a
unit equals zero. I need to create zero data for the missing units.

For example:
day<-(c(rep(4,8),rep(6,8)))
unit<-c(seq(1,8,1),seq(2,16,2))
value<-floor(rnorm(16,25,10))
x<-cbind(day,unit,value)

Thanks for the actual reproducible example.

x
? ? day unit value
?[1,] ? 4 ? ?1 ? ?19
?[2,] ? 4 ? ?2 ? ?15
?[3,] ? 4 ? ?3 ? ?16
?[4,] ? 4 ? ?4 ? ?20
?[5,] ? 4 ? ?5 ? ?17
?[6,] ? 4 ? ?6 ? ?15
?[7,] ? 4 ? ?7 ? ?14
?[8,] ? 4 ? ?8 ? ?29
?[9,] ? 6 ? ?2 ? ?18
[10,] ? 6 ? ?4 ? ?22
[11,] ? 6 ? ?6 ? ?27
[12,] ? 6 ? ?8 ? ?16
[13,] ? 6 ? 10 ? ?45
[14,] ? 6 ? 12 ? ?36
[15,] ? 6 ? 14 ? ?34
[16,] ? 6 ? 16 ? ?13

Lets say the stream has 16 units. For each day, I want to fill in rows
for
any missing units (e.g., units 9-16 for day 4, the odd numbered units on
day
6) with values of zero.

I could not figure out what you wanted precisely. If "day" is the row
designator, and you want values by 'unit' and 'day' with zeros for the
missing, then that is exactly what `xtab` delivers:

xtabs(value ~ day+unit, data=x)

? unit
day ?1 ?2 ?3 ?4 ?5 ?6 ?7 ?8 10 12 14 16
?4 25 34 ?3 25 38 18 19 33 ?0 ?0 ?0 ?0
?6 ?0 22 ?0 42 ?0 37 ?0 ?4 12 31 17 28

You cannot get much more concise than that.

--
david.

Here's one option, though it may not be terribly concise:

all.samples <- expand.grid(day=unique(x[,"day"]), unit=1:16)
all.samples <- all.samples[order(all.samples[,"day"],
all.samples[,"unit"]),]
x.final <- merge(x, all.samples, all.y=TRUE)
x.final[is.na(x.final[,"value"]), "value"] <- 0

Sarah

Does anyone know a relatively concise way to do this?
Thank you.

? ? ? [[alternative HTML version deleted]]

--
Sarah Goslee
http://www.functionaldiversity.org

David Winsemius, MD
West Hartford, CT

Prompted by David's xtabs() suggestion, one way to do what I think the
OP wants is to
* define day and unit as factors whose levels comprise the full range
of desired values;
* use xtabs();
* return the result as a data frame.
Something like

x <- data.frame( day = factor(rep(c(4, 6), each = 8), levels = 4:6),
                unit = factor(c(1:8, seq(2,16,2)), levels = 1:16),
                value = floor(rnorm(16,25,10)) )
as.data.frame(with(x, xtabs(value ~ unit + day)))