Hi R Users,
I do have very big data sets and wanted to run some of the analyses many
times with randomization (1000 times).
I have done the analysis using an example data but it need to be done with
randomized data (1000 times). I am doing manually for 10000 times but
taking so much time, I wonder whether it is possible to perform the
analysis with creating a loop for many replicated datasets? The code and
the example data sets are attached.
I will be very grateful if someone help me to create the loop for the
following example data and the analyses.
I appreciate your help.
MW
#####
dat1<-structure(list(RegionA = structure(c(1L, 1L, 2L, 3L, 3L, 4L, 5L, 5L,
6L, 6L, 6L, 6L, 6L, 6L, 6L, 6L, 6L, 6L, 6L), .Label = c("Ra", "Rb", "Rc",
"Rd", "Re", "Rf"), class = "factor"), site = structure(c(1L, 12L, 13L, 14L,
15L, 16L, 17L, 18L, 19L, 2L, 3L, 4L, 5L, 6L, 7L, 8L, 9L, 10L, 11L), .Label
= c("s1", "s10", "s11", "s12", "s13", "s14", "s15", "s16", "s17", "s18",
"s19", "s2", "s3", "s4", "s5", "s6", "s7", "s8", "s9"), class = "factor"),
temp = c(23L, 21L, 10L, 15L, 16L, 8L, 13L, 1L, 23L, 19L, 25L, 19L, 12L, 16L,
19L, 21L, 12L, 5L, 7L), group = structure(c(1L, 1L, 1L, 2L, 2L, 2L,
2L, 2L, 2L, 2L, 3L, 3L, 3L, 3L, 3L, 3L, 3L, 3L, 3L), .Label = c("A", "B",
"C"), class = "factor")), .Names = c("RegionA", "site", "temp", "group"),
class = "data.frame", row.names = c(NA, -19L))
head(dat1)
dat2<-structure(list(group = structure(1:3, .Label = c("A", "B", "C"
), class = "factor"), totalP = c(250L, 375L, 180L), sampled = c(25L,
37L, 27L)), .Names = c("group", "total.pop", "sampled.pop"), class =
"data.frame", row.names = c(NA,
-3L))
##
idx <- 1:nrow(dat1)
lll <- split(idx, dat1$group)
##########################
#Replication 1 create a resampled data
############################
Replication1<-dat1[unlist(lapply(lll, sample, rep=TRUE)),]
Summary.Rep1<-ddply(Replication1, c("group"), summarise,
N = length(group),
mean = mean(temp, na.rm=TRUE),
sd = sd(temp),
se = sd / sqrt(N),
variance=sd^2
)
#merge two datasets (dat1 and dat2)
Rep1<-merge(Summary.Rep1, dat2, by="group")
#calclate adjusted mean. variance
Rep1$adj.mean<-(Rep1$total.pop*Rep1$mean)/sum(Rep1$total.pop)
Rep1$adj.var<-(Rep1$variance)/(Rep1$sampled.pop/(1-(Rep1$sampled.pop/Rep1$
total.pop)))
Rep1$over.adj.var<-(Rep1$total.pop/sum(Rep1$total.pop))^2*Rep1$adj.var
Rep1$total<-Rep1$adj.mean*(Rep1$total.pop)
##
Estimated.TotalTemp<-sum(Rep1$adj.mean)*sum(Rep1$total.pop)
Estimated.totalvar<-sum(Rep1$adj.var)
Estimated.SE<-sqrt(Estimated.totalvar)*sum(Rep1$total.pop)
RESULTS.R1<-data.frame(Estimated.TotalTemp, SE=Estimated.SE)
RESULTS.R1
##########################
#Replication 2 create a resampled data
############################
Replication2<-dat1[unlist(lapply(lll, sample, rep=TRUE)),]
Summary.Rep2<-ddply(Replication2, c("group"), summarise,
N = length(group),
mean = mean(temp, na.rm=TRUE),
sd = sd(temp),
se = sd / sqrt(N),
variance=sd^2
)
#merge two datasets
Rep1<-merge(Summary.Rep2, dat2, by="group")
#calclate adjusted mean. variance
Rep2$adj.mean<-(Rep2$total.pop*Rep2$mean)/sum(Rep2$total.pop)
Rep2$adj.var<-(Rep2$variance)/(Rep2$sampled.pop/(1-(Rep2$sampled.pop/Rep2$
total.pop)))
Rep2$over.adj.var<-(Rep2$total.pop/sum(Rep2$total.pop))^2*Rep2$adj.var
Rep2$total<-Rep2$adj.mean*(Rep2$total.pop)
##
Estimated.TotalTemp<-sum(Rep2$adj.mean)*sum(Rep2$total.pop)
Estimated.totalvar<-sum(Rep2$adj.var)
Estimated.SE<-sqrt(Estimated.totalvar)*sum(Rep2$total.pop)
RESULTS.R2<-data.frame(Estimated.TotalTemp, SE=Estimated.SE)
##############################
#combined all results from 1000 runs
ALL.Results(Restult.R1, Result.R2....)
[[alternative HTML version deleted]]
______________________________________________
R-help@r-project.org mailing list -- To UNSUBSCRIBE and more, see
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.
