Simulating t-test p-values using a for loop - r

For this project I am required to use an R script to simulate the effectiveness of the t-test. I must use a for loop will be used to carry out the following 2000 times:
Would the loop look something like this
i <- 1
for (i <= 2001) {
x <-rf(5,df1=5,df2=10)
b <- df2
p.value <-t.test(x,mu=(b/(b-2))$p.value
i <- i+1
}

In the way you wrote it, it would be a "while" loop.
For loops in R have the following syntax:
for (i in 1:2000) {
df1 <- 5
df2 <- 10
x <-rf(5, df1=df1, df2=df2)
b <- df2
p.value <- t.test(x, mu=(b/(b-2)))$p.value
}
Additionally, it might be more efficient to employ an "apply" construct, for example with replicate, and include the df as function arguments:
get.p.value <- function(df1, df2) {
x <- rf(5, df1=df1, df2=df2)
p.value <- t.test(x, mu=(df2/(df2-2)))$p.value
}
replicate (2000, get.p.value(df1 = 5, df2 = 10))
This is not always true, but it simplifies the recovery of the p.values.

Related

R - Cleanest way to run statistical test on every permutation of multiple populations

I have three populations stored as individual vectors. I need to run a statistical test (wilcoxon, if it matters) on each pair of these three populations.
I want to input three vectors into some block of code and get as output a vector of 6 p-values (one p-value is the result of one test and is a double).
I have a method that works but I am new to R and from what I've been reading I feel like there should be a better way, possibly involving storing the vectors as a data frame and using vectorization, to write this code.
Here is the code I have:
library(arrangements)
runAllTests <- function(pop1,pop2,pop3) {
populations <- list(pop1=pop1,pop2=pop2,pop3=pop3)
colLabels <- c("pop1", "pop2", "pop3")
#This line makes a data frame where each column is a pair of labels
perms <- data.frame(t(permutations(colLabels,2)))
pvals <- vector()
#This for loop gets each column of that data frame
for (pair in perms[,]) {
pair <- as.vector(pair)
p1 <- as.numeric(unlist(populations[pair[1]]))
p2 <- as.numeric(unlist(populations[pair[2]]))
pvals <- append(pvals, wilcox.test(p1, p2,alternative=c("less"))$p.value)
}
return(pvals)
}
What is a more R appropriate way to write this code?
Note: Generating populations and comparing them all to each other is a common enough thing (and tricky enough to code) that I think this question will apply to more people than myself.
EDIT: I forgot that my actual populations are of different sizes. This means I cannot make a data frame out of the vectors (as far as I know). I can make a list of vectors though. I have updated my code with a version that works.
Yes, this is indeed common; indeed so common that R has a built-in function for exactly this scenario: pairwise.table.
p <- list(pop1, pop2, pop3)
pairwise.table(function(i, j) {
wilcox.test(p[[i]], p[[j]])$p.value
}, 1:3)
There are also specific versions for t tests, proportion tests, and Wilcoxon tests; here's an example using pairwise.wilcox.test.
p <- list(pop1, pop2, pop3)
d <- data.frame(x=unlist(p), g=rep(seq_along(p), sapply(p, length)))
with(d, pairwise.wilcox.test(x, g))
Also, make sure you look into the p.adjust.method parameter to correctly adjust for multiple comparisons.
Per your comments, you're interested in tests where the order matters; that's really hard to imagine (and isn't true for the Wilcoxon test you mentioned) but still...
This is the pairwise.table function, edited to do tests in both directions.
pairwise.table.all <- function (compare.levels, level.names, p.adjust.method) {
ix <- setNames(seq_along(level.names), level.names)
pp <- outer(ix, ix, function(ivec, jvec)
sapply(seq_along(ivec), function(k) {
i <- ivec[k]; j <- jvec[k]
if (i != j) compare.levels(i, j) else NA }))
pp[] <- p.adjust(pp[], p.adjust.method)
pp
}
This is a version of pairwise.wilcox.test which uses the above function, and also runs on a list of vectors, instead of a data frame in long format.
pairwise.lazerbeam.test <- function(dat, p.adjust.method=p.adjust.methods) {
p.adjust.method <- match.arg(p.adjust.method)
level.names <- if(!is.null(names(dat))) names(dat) else seq_along(dat)
PVAL <- pairwise.table.all(function(i, j) {
wilcox.test(dat[[i]], dat[[j]])$p.value
}, level.names, p.adjust.method = p.adjust.method)
ans <- list(method = "Lazerbeam's special method",
data.name = paste(level.names, collapse=", "),
p.value = PVAL, p.adjust.method = p.adjust.method)
class(ans) <- "pairwise.htest"
ans
}
Output, both before and after tidying, looks like this:
> p <- list(a=1:5, b=2:8, c=10:16)
> out <- pairwise.lazerbeam.test(p)
> out
Pairwise comparisons using Lazerbeams special method
data: a, b, c
a b c
a - 0.2821 0.0101
b 0.2821 - 0.0035
c 0.0101 0.0035 -
P value adjustment method: holm
> pairwise.lazerbeam.test(p) %>% broom::tidy()
# A tibble: 6 x 3
group1 group2 p.value
<chr> <chr> <dbl>
1 b a 0.282
2 c a 0.0101
3 a b 0.282
4 c b 0.00350
5 a c 0.0101
6 b c 0.00350
Here is an example of one approach that uses combn() which has a function argument that can be used to easily apply wilcox.test() to all variable combinations.
set.seed(234)
# Create dummy data
df <- data.frame(replicate(3, sample(1:5, 100, replace = TRUE)))
# Apply wilcox.test to all combinations of variables in data frame.
res <- combn(names(df), 2, function(x) list(data = c(paste(x[1], x[2])), p = wilcox.test(x = df[[x[1]]], y = df[[x[2]]])$p.value), simplify = FALSE)
# Bind results
do.call(rbind, res)
data p
[1,] "X1 X2" 0.45282
[2,] "X1 X3" 0.06095539
[3,] "X2 X3" 0.3162251

condition in loop in R

I have a relative simple question for which I was not able to apply solutions I have found on the internet. Let's say we have:
set.seed(20)
data <- data.frame(month = rep(month.name, 25),
a = rnorm(300, 0, 1), b = runif(300, 0, 7.2))
I want to calculate using a loop the f-test for variance between columns a and b for each month in month. This I done by using:
# create some empty vectors to fill in later
pval <- as.double()
ftest <- as.double()
month <- as.character()
# looping through the months
for (i in unique(data$month)){
print(i)
# sh.1 <- shapiro.test(data$a[data$month==i])
# sh.1[2] > 0.05 # apply log if it's smaller than 0.05
# sh.2 <- shapiro.test(data$b[data$month==i])
# sh.2[2] > 0.05 # apply log if it's smaller than 0.05
var.t <- var.test(data$a[data$month==i], data$b[data$month==i])
f <- round(var.t[[1]],2)
p <- round(var.t$p.value,2)
ftest <- append(ftest, f)
pval <- append(pval, p)
month <- append(month, i)
}
However, as far as I know, f-test is very sensitive to normal distribution. Therefore, I am planning to use a condition into loop where in case that p-value of shapiro test is smaller than 0.05 a log transformation for the data will be required; then it will be used into f-test.
Normally, I would to this with an ifelse condition but I am not very sure how to use it here.
Any help here please?
I believe the code below does what you want. It uses *apply loops, not for loops in order to make the code more readable (I think).
First I will recreate the data and make sure column a is all positive.
set.seed(20)
data <- data.frame(month = rep(month.name, 25),
a = rnorm(300, 0, 1), b = runif(300, 0, 7.2))
data$a <- abs(data$a)
Now, instead of looping through unique values of month, I split the data.frame by that variable. Like this each of the df's in the resulting list sp already is a df of all rows of each month.
sp <- split(data, data$month)
sp <- sp[order(order(month.name))]
It's here that the data are log transformed if necessary.
sp <- lapply(sp, function(DF){
if(shapiro.test(DF[["a"]])$p.value < 0.05) DF[["a"]] <- log(DF[["a"]])
if(shapiro.test(DF[["b"]])$p.value < 0.05) DF[["b"]] <- log(DF[["b"]])
DF
})
And lapply the test you want, var.test, to all of these data.frames.
vartest_list <- lapply(sp, function(DF){
var.t <- var.test(DF[["a"]], DF[["b"]])
list(f = var.t[[1]],
p.value = var.t$p.value,
month = as.character(DF[["month"]][1]))
})
Finally, it is a simple matter of applying the extraction function [[ to the tests' results. This works because hypothesis tests functions in R return objects of class "htest" that are nothing else but lists. The last of the extraction loops is commented out.
ftest <- sapply(vartest_list, '[[', 'f')
pval <- sapply(vartest_list, '[[', 'p.value')
#month <- sapply(vartest_list, '[[', 'month')

How to make combination of sample and lapply reproducible?

The followings are functions for bootstrapping, but how can I make the result reproducible? I tried set.seed() but that does not work because the every time lapply calls function boot.lm.vector, the function just produced one simulated set and calculated coefficients once. Is there any thing in R that can function like a seed list? or any other way to make the result reproducible?
boot.lm.vector <- function(index, inputData) {
d <- inputData[sample.int(nrow(inputData), replace = T),]
a <- ncol(inputData)-1
X <- d[, 1:a]
y <- d[, a+1]
solve(crossprod(X), crossprod(X,y))
}
rtest <- lapply(1:10000, fun = boot.lm.vector, inputData = boot_set)
rtestdf <- plyr::ldply(rtest)
If you set the seed using an index inside your function, you should be able to reproduce it. Dummy boot.lm.vector function below:
## samples 1 item from inputData
boot.lm.vector <- function(index, inputData) {
set.seed(index)
return(sample(inputData, 1))
}
## iterating 5 times: use lapply as per your requirement
test <- sapply(1:5, FUN = boot.lm.vector, inputData = 1:10)
test
[1] 3 2 2 6 3 # reproducible result

R: Replacing a for-loop with an apply function

I managed to apply a linear regression for each subject of my data frame and paste the values into a new dataframe using a for-loop. However, I think there should be a more readable way of achieving my result using an apply function, but all my attempts fail. This is how I do it:
numberOfFiles <- length(resultsHick$subject)
intslop <- data.frame(matrix(0,numberOfFiles,4))
intslop <- rename(intslop,
subject = X1,
intercept = X2,
slope = X3,
Rsquare = X4)
cond <- c(0:3)
allSubjects <- resultsHick$subject
for (i in allSubjects)
{intslop[i,1] <- i
yvalues <- t(subset(resultsHick,
subject == i,
select = c(H0meanRT, H1meanRT, H2meanRT, H258meanRT)))
fit <- lm(yvalues ~ cond)
intercept <- fit$coefficients[1]
slope <- fit$coefficients[2]
rsquared <- summary(fit)$r.squared
intslop[i,2] <- intercept
intslop[i,3] <- slope
intslop[i,4] <- rsquared
}
The result should look the same as
> head(intslop)
subject intercept slope Rsquare
1 1 221.3555 54.98290 0.9871209
2 2 259.4947 66.33344 0.9781499
3 3 227.8693 47.28699 0.9537868
4 4 257.7355 80.71935 0.9729132
5 5 197.4659 49.57882 0.9730409
6 6 339.1649 61.63161 0.8213179
...
Does anybody know a more readable way of writing this code using an apply function?
One common pattern I use to replace for loops that aggregate data.frames is:
do.call(
rbind,
lapply(1:numberOfDataFrames,
FUN = function(i) {
print(paste("Processing index:", i)) # helpful to see how slow/fast
temp_df <- do_some_work[i]
temp_df$intercept <- 1, etc.
return(temp_df) # key is to return a data.frame for each index.
}
)
)

creating a function for processing my dataframe calculations

I am doing systematic calculations for my created dataframe. I have the code for the calculations but I would like to:
1) Wite it as a function and calling it for the dataframe I created.
2) reset the calculations for next ID in the dataframe.
I would appreciate your help and advice on this.
The dataframe is created in R using the following code:
#Create a dataframe
dosetimes <- c(0,6,12,18)
df <- data.frame("ID"=1,"TIME"=sort(unique(c(seq(0,30,1),dosetimes))),"AMT"=0,"A1"=NA,"WT"=NA)
doserows <- subset(df, TIME%in%dosetimes)
doserows$AMT[doserows$TIME==dosetimes[1]] <- 100
doserows$AMT[doserows$TIME==dosetimes[2]] <- 100
doserows$AMT[doserows$TIME==dosetimes[3]] <- 100
doserows$AMT[doserows$TIME==dosetimes[4]] <- 100
#Add back dose information
df <- rbind(df,doserows)
df <- df[order(df$TIME,-df$AMT),]
df <- subset(df, (TIME==0 & AMT==0)==F)
df$A1[(df$TIME==0)] <- df$AMT[(df$TIME ==0)]
#Time-dependent covariate
df$WT <- 70
df$WT[df$TIME >= 12] <- 120
#The calculations are done in a for-loop. Here is the code for it:
#values needed for the calculation
C <- 2
V <- 10
k <- C/V
#I would like this part to be written as a function
for(i in 2:nrow(df))
{
t <- df$TIME[i]-df$TIME[i-1]
A1last <- df$A1[i-1]
df$A1[i] = df$AMT[i]+ A1last*exp(-t*k)
}
head(df)
plot(A1~TIME, data=df, type="b", col="blue", ylim=c(0,150))
The other thing is that the previous code assumes the subject ID=1 for all time points. If subject ID=2 when the WT (weight) changes to 120. How can I reset the calculations and make it automated for all subject IDs in the dataframe? In this case the original dataframe would be like this:
#code:
rm(list=ls(all=TRUE))
dosetimes <- c(0,6,12,18)
df <- data.frame("ID"=1,"TIME"=sort(unique(c(seq(0,30,1),dosetimes))),"AMT"=0,"A1"=NA,"WT"=NA)
doserows <- subset(df, TIME%in%dosetimes)
doserows$AMT[doserows$TIME==dosetimes[1]] <- 100
doserows$AMT[doserows$TIME==dosetimes[2]] <- 100
doserows$AMT[doserows$TIME==dosetimes[3]] <- 100
doserows$AMT[doserows$TIME==dosetimes[4]] <- 100
df <- rbind(df,doserows)
df <- df[order(df$TIME,-df$AMT),]
df <- subset(df, (TIME==0 & AMT==0)==F)
df$A1[(df$TIME==0)] <- df$AMT[(df$TIME ==0)]
df$WT <- 70
df$WT[df$TIME >= 12] <- 120
df$ID[(df$WT>=120)==T] <- 2
df$TIME[df$ID==2] <- c(seq(0,20,1))
Thank you in advance!
In general, when doing calculations on different subject's data, I like to split the dataframe by ID, pass the vector of individual subject data into a for loop, do all the calculations, build a vector containing all the newly calculated data and then collapse the resultant and return the dataframe with all the numbers you want. This allows for a lot of control over what you do for each subject
subjects = split(df, df$ID)
forResults = vector("list", length=length(subjects))
# initialize these constants
C <- 2
V <- 10
k <- C/V
myFunc = function(data, resultsArray){
for(k in seq_along(subjects)){
df = subjects[[k]]
df$A1 = 100 # I assume this should be 100 for t=0 for each subject?
# you could vectorize this nested for loop..
for(i in 2:nrow(df)) {
t <- df$TIME[i]-df$TIME[i-1]
A1last <- df$A1[i-1]
df$A1[i] = df$AMT[i]+ A1last*exp(-t*k)
}
head(df)
# you can add all sorts of other calculations you want to do on each subject's data
# when you're done doing calculations, put the resultant into
# the resultsArray and we'll rebuild the dataframe with all the new variables
resultsArray[[k]] = df
# if you're not using RStudio, then you want to use dev.new() to instantiate a new plot canvas
# dev.new() # dont need this if you're using RStudio (which doesnt allow multiple plots open)
plot(A1~TIME, data=df, type="b", col="blue", ylim=c(0,150))
}
# collapse the results vector into a dataframe
resultsDF = do.call(rbind, resultsArray)
return(resultsDF)
}
results = myFunc(subjects, forResults)
Do you want this:
ddf <- data.frame("ID"=1,"TIME"=sort(unique(c(seq(0,30,1),dosetimes))),"AMT"=0,"A1"=NA,"WT"=NA)
myfn = function(df){
dosetimes <- c(0,6,12,18)
doserows <- subset(df, TIME%in%dosetimes)
doserows$AMT[doserows$TIME==dosetimes[1]] <- 100
doserows$AMT[doserows$TIME==dosetimes[2]] <- 100
doserows$AMT[doserows$TIME==dosetimes[3]] <- 100
doserows$AMT[doserows$TIME==dosetimes[4]] <- 100
#Add back dose information
df <- rbind(df,doserows)
df <- df[order(df$TIME,-df$AMT),]
df <- subset(df, (TIME==0 & AMT==0)==F)
df$A1[(df$TIME==0)] <- df$AMT[(df$TIME ==0)]
#Time-dependent covariate
df$WT <- 70
df$WT[df$TIME >= 12] <- 120
#The calculations are done in a for-loop. Here is the code for it:
#values needed for the calculation
C <- 2
V <- 10
k <- C/V
#I would like this part to be written as a function
for(i in 2:nrow(df))
{
t <- df$TIME[i]-df$TIME[i-1]
A1last <- df$A1[i-1]
df$A1[i] = df$AMT[i]+ A1last*exp(-t*k)
}
head(df)
plot(A1~TIME, data=df, type="b", col="blue", ylim=c(0,150))
}
myfn(ddf)
For multiple calls:
for(i in 1:N) {
myfn(ddf[ddf$ID==i,])
readline(prompt="Press <Enter> to continue...")
}

Resources