repeated measures anova: length of 'dimnames' [1] not equal to array extent - r

I conducted the repeated measures anova, that's my code, it's simple operation and i always did it quickly.
Link to mydata in .csv format
library(car)
vivo4 <- read.csv("vivo1.csv",sep=";",dec=",")
ageLevels <- c(1, 2,3,4,5,6,7,8,9,10,12)
ageFactor <- as.factor(ageLevels)
ageFrame <- data.frame(ageFactor)
measures <- function(data = vivo4, n = 4) { #n=4 is 4 variables
## Editor comment:
## correct way to initialize a list, don't use "list(n)"
## you can compare what you get from "list(4)" and "vector ("list", length = 4)"
## lmo's comment: don't use "list" for your variable name (may mask R function "list")
## I have corrected it as "Mylist"
Mylist <- vector("list", length = n)
for(i in 0:3) {Mylist[[i+1]] <- as.matrix(cbind(data[, 12*i + 1:12])) # 12 visits
}
Mylist
}
measures_list <- measures()
models <- lapply(
measures_list, function(x) {
ageModel <- lm(x ~ 1)
Anova.mlm (ageModel, idata = ageFrame, idesign = ~ageFactor)
} )
models #View the result
but i got the error
Error in `rownames<-`(`*tmp*`, value = colnames(B)) :
length of 'dimnames' [1] not equal to array extent
I have read many answers and can't understand whats wrong, i need supervision.


You have 12 levels in your dataset,but in agelevels, you indicated only 11 levels.
i.e. you forgot indicate 14
ageLevels <- c(1, 2,3,4,5,6,7,8,9,10,12,14)

Related

I am having an issue adding vectors to a list [duplicate]

This question already has an answer here:
What is the difference between [ ] and [[ ]] in R? [duplicate]
(1 answer)
Closed 1 year ago.
I have a list of matrices constructed by the following loops:
# Set up Row and Column Names for prediction coefficients.
rows = c("Intercept", "actsBreaks0", "actsBreaks1","actsBreaks2","actsBreaks3","actsBreaks4","actsBreaks5","actsBreaks6",
"actsBreaks7","actsBreaks8","actsBreaks9","tBreaks0","tBreaks1","tBreaks2","tBreaks3", "unitBreaks0", "unitBreaks1",
"unitBreaks2","unitBreaks3", "covgBreaks0","covgBreaks1","covgBreaks2","covgBreaks3","covgBreaks4","covgBreaks5",
"covgBreaks6","yearBreaks2016","yearBreaks2015","yearBreaks2014","yearBreaks2013","yearBreaks2011",
"yearBreaks2010","yearBreaks2009","yearBreaks2008","yearBreaks2007","yearBreaks2006","yearBreaks2005",
"yearBreaks2004","yearBreaks2003","yearBreaks2002","yearBreaks2001","yearBreaks2000","yearBreaks1999",
"yearBreaks1998","plugBump0","plugBump1","plugBump2","plugBump3")
cols = c("Value")
# Build Matrix for dummy coefficient values.
matrix1 <- matrix(c(1:48), nrow = 48, ncol = 1, byrow = TRUE, dimnames = list(rows,cols))
matrix1
# Extract each variable type into own matrix (i.e. all "actsBreaks{x}")
#
Beta_names <- list()
betabreaks <- unique(gsub("[0-9]*", "", rows))
for (bc in betabreaks)
{
Breaks <- grep(paste0(bc, "[0-9]*"), rows)
Beta_names[[bc]] <- matrix1[Breaks, ,drop = FALSE]
Beta_names[[bc]] <- data.matrix(unlist(Beta_names[[bc]])) #, byrow = TRUE)
}
# Set up matrices for excluded/test data
one_column <-c(1,1,2,3,3,3,4,4,4,4,4,5,6,9,9,8,7,5,4,7,7,8,2,0,10)
two_column <-c(1,1,2,3,3,3,4,4,4,4,4,5,6,9,9,8,7,5,4,7,7,8,3,0,10)
three_column <-c(1,1,2,3,3,3,4,4,4,4,4,5,6,9,9,8,7,5,4,7,7,8,4,10,0)
four_column <-c(1,1,2,3,3,3,4,4,4,4,4,5,6,9,9,8,7,5,4,7,7,8,5,0,10)
five_column <-c(1,1,2,3,3,3,4,4,4,4,4,5,6,9,9,8,7,5,4,7,7,8,6,0,10)
six_column <-c(1,1,2,3,3,3,4,4,4,4,4,5,6,9,9,8,7,5,4,7,7,8,7,0,10)
seven_column <-c(1,1,2,3,3,3,4,4,4,4,4,5,6,9,9,8,7,5,4,7,7,8,8,0,10)
eight_column <-c(1,1,2,3,3,3,4,4,4,4,4,5,6,9,9,8,7,5,4,7,7,8,9,0,10)
nine_column <-c(1,1,2,3,3,3,4,4,4,4,4,5,6,9,9,8,7,5,4,7,7,8,1,0,10)
ten_column <-c(1,1,2,3,3,3,4,4,4,4,4,5,6,9,9,8,7,5,4,7,7,8,0,0,10)
DF1 <- data.frame (one_column ,two_column ,three_column ,
four_column ,five_column ,six_column ,
seven_column ,eight_column ,nine_column ,
ten_column )
paralength <- 5
Xnames <- list()
datindc <- 1
while ( datindc <= paralength )
{
Xbreaks <- factor(DF1[[datindc]],levels=sort(unique.default(DF1[[datindc]]),decreasing=FALSE))
Xnames[[datindc]] <- data.frame(model.matrix(~Xbreaks -1), stringsAsFactors = FALSE)
datindc <- datindc + 1
}
#
Xlngth <- length(Xnames)
BtaXind <- 1
BetaiXi <- list()
while ( BtaXind <= Xlngth )
{
BetaiXi[[BtaXind]] <- (Beta_names[[BtaXind + 1]] * Xnames[[BtaXind]])
BtaXind <- (BtaXind + 1)
}
I need to add each of those matrices' rows to each other, which I am trying to do by turning each matrix into a vector
BiXilngth <- length(BetaiXi)
BetaiXiTr <- list()
BtaiXiTrd <- 1
while (BtaiXiTrd <= BiXilngth)
{
Var1 <- c(t(BetaiXi[[BtaiXiTrd]]))
BetaiXiTr[BtaiXiTrd] <- Var1
BtaiXiTrd <- BtaiXiTrd + 1
}
and adding the vectors, effectively transposing the matrices. However, when I tried to convert the first matrix BetaiXi[[1]] to a vector and add it to the list with this command BetaiXiTr[BtaiXiTrd] <- c(t(BetaiXi[[BtaiXiTrd]])) I got the following message:
Warning message:
In BetaiXiTr[BtaiXiTrd] <- c(t(BetaiXi[[BtaiXiTrd]])) :
number of items to replace is not a multiple of replacement length
I then tried using unlist():
> BetaiXiTr[BtaiXiTrd] <-unlist(c(t(BetaiXi[[1]])))
Warning message:
In BetaiXiTr[BtaiXiTrd] <- unlist(c(t(BetaiXi[[1]]))) :
number of items to replace is not a multiple of replacement length
with the same result. Finally, I tried assigning the first vector to a variable > Var1 <- c(t(BetaiXi[[BtaiXiTrd]])) and assigning that vector to the list > BetaiXiTr[BtaiXiTrd] <- Var1 with, yet again, the same warning:
Warning message:
In BetaiXiTr[BtaiXiTrd] <- Var1 :
number of items to replace is not a multiple of replacement length
I searched for the warning message to determine what exactly I was being warned of but ended being more confused. Most reproduce or encountered the error message by trying to replace a vector of so many elements with a vector of fewer, while (to my understanding) I am simply trying to add a vector to a list. Am I going about this the incorrect way?

I was using [ ] and [ [ ] ] incorrectly in BetaiXiTr[BtaiXiTrd]. It needs to be BetaiXiTr[[BtaiXiTrd]]and that allows the vectors to be added.

Condition to check values ​between lists and add to new list in R

If the last value of each sublist in the list ListResiduals (e.g: OptionAOptionD) is > than the value with the corresponding name in the ListSigma (e.g: OptionAOptionD), it adds the name (e.g: OptionAOptionD) to the Watchlist list.
In the last line of the code I put "> 5" just for the example work, it's the "> 5" that I want to replace in the condition that I mentioned in the previous paragraph.
DF <- data.frame("OptionA" = sample(1:100, 50),
"OptionB" = sample(1:100, 50),
"OptionC" = sample(1:100, 50),
"OptionD" = sample(1:100, 50))
#Unfolding options and creating DF
UnFolding <- data.frame(
First = as.vector(sapply(names(DF[]), function(x)
sapply(names(DF[]), function(y)
paste0(x)))),
Second = as.vector(sapply(names(DF[]), function(x)
sapply(names(DF[]), function(y)
paste0(y)))))
#Deleting lines with the same names
UnFolding <-
UnFolding[UnFolding$First != UnFolding$Second, ]
#Creating list with dependent and independent variables
LMList <- apply(UnFolding, 1, function(x)
as.formula(paste(x[1], "~", x[2])))
#Change list data to variable names
names(LMList) <- substring(lapply(LMList, paste, collapse = ""), 2)
#Linear regression - lm()
LMListRegression <- lapply(LMList, function(x) {
eval(call("lm", formula = x, data = DF))
})
#Residuals
ListResiduals <- lapply(LMListRegression, residuals)
#Sigma
ListSigma <- lapply(LMListRegression, function(x) {
sigma(x)*2
})
#Watchlist
Watchlist <- as.list(unlist(lapply(ListResiduals,
function(x) names(x)[1][tail(x, 1) > 5])))

I would gravitate towards converting your Simga and Residual values to a vector and compare the vectors. You could also use a data.frame approach to be sure the order of your lists/vectors doesn't change.
# create a vector with the last value from the Residuals list.
last_residual <- sapply(ListResiduals, `[`, 50)
names(last_residual) <- substr(names(last_residual), 1, stop = -4)
# Using sapply() rather than lapply, will return a named vector
sigma_vector <- sapply(LMListRegression, function(x) {
sigma(x)*2
})
Watchlist <- sigma_vector[last_residual > sigma_vector]
Watchlist
# named numeric(0)
In your example, it returns an empty named vector because no values meet your condition
max(last_residual)
# [1] 31.70949
min(sigma_vector)
# [1] 52.93234
# To demonstrate that it works, let's devide sigma by 2 so that at least some values will pass
half_sigma <- sigma_vector/2
Watchlist2 <- sigma_vector[last_residual > half_sigma]
Watchlist2
# OptionDOptionA OptionDOptionB OptionDOptionC
# 54.52411 57.09503 56.79341

R for loop to calculate wilcox.test

I am trying to write a code that would automatically calculate Wilcoxon test p-value for several comparisons.
Data used: 2 data sets with the same information representing two groups of participants completed the same 5 tasks which means that the each table contains 5 columns (tasks) and X rows with tasks scores.
data_17_18_G2 # first data set (in data.table format)
data_18_20_G2 # second data set (in data.table format)
Both data sets have identical names of column which are to be used in the W-test the next way:
wilcox.test(Group1Task1, Group2Task1, paired = F)
wilcox.test(Group1Task2, Group2Task2, paired = F)
and so on.
The inputs (e.g., Grou1Task1) are two vectors of task scores (the first one will be from data_17_18_G2 and the other one from data_18_20_G2
Desired output: a data table with a column of p-values
The problem I faced is that no matter how I manipulated the val1 and val2 empty objects, in the second and the third lines the right size "as.numeric(unlist(data_17_18_G2[, ..i]))" gives a correct output (a numeric vector) but it's left size "val1[i]" always returns only one value from the vector. That gave me the idea that the main problem appeared on the step of creating an empty vector, however, I wasn't able to solve it.
Empty objects:
result <- data.table(matrix(ncol=2))
val1 <- as.numeric() # here I also tried functions "numeric" and "vector"
val2 <- as.numeric()
res <- vector(mode = "list", length = 7)
For loop
for (i in 1:5) {
val1[i] <- as.numeric(unlist(data_17_18_G2[ , ..i]))
val2[i] <- as.numeric(unlist(data_18_20_G2[ , ..i]))
res[i] <- wilcox.test(val1[i], val2[i], paired = F)
result[i, 1] <- i
result[i, 2] <- res$p.value
}
Output:
Error in `[<-.data.table`(`*tmp*`, i, 2, value = NULL) :
When deleting columns, i should not be provided
1: В val1[i] <- as.numeric(unlist(data_17_18_G2[, ..i])) :
number of items to replace is not a multiple of replacement length
2: В val2[i] <- as.numeric(unlist(data_18_20_G2[, ..i])) :
number of items to replace is not a multiple of replacement length
3: В res[i] <- wilcox.test(val1[i], val2[i], paired = F) :
number of items to replace is not a multiple of replacement length
Alternative:
I changed the second and the third lines
for (i in 1:5) {
val1[i] <- as.numeric(data_17_18_G2[ , ..i])
val2[i] <- as.numeric(data_18_20_G2[ , ..i])
res[i] <- wilcox.test(val1[i], val2[i], paired = F)
result[i, 1] <- i
result[i, 2] <- res$p.value
}
And got this
Error in as.numeric(data_17_18_G2[, ..i]) :
(list) object cannot be coerced to type 'double'
which means that the function wilcox.test cannot interpret this type of input.
How can I improve the code so that I get a data table of p-values?

There would appear to be some bugs in the code. I have rewritten the code using the cars dataset as a example.
## use the cars dataset as a example (change with appropriate data)
data(cars)
data_17_18_G2 <- as.data.table(cars)
data_18_20_G2 <- data_17_18_G2[,2:1]
## Fixed code
result <- data.table(matrix(as.numeric(), nrow=ncol(data_17_18_G2), ncol=2))
val1 <- as.numeric()
val2 <- as.numeric()
res <- vector(mode = "list", length = 7)
for (i in 1:ncol(data_17_18_G2)) {
val1 <- as.numeric(unlist(data_17_18_G2[ , ..i]))
val2 <- as.numeric(unlist(data_18_20_G2[ , ..i]))
res[[i]] <- wilcox.test(val1, val2, paired = F)
result[i, 1] <- as.numeric(i)
result[i, 2] <- as.numeric(res[[i]]$p.value)
}
Hope this gives you the output you are after.

Why does function boot return values of different types than those from the statistic used inside boot?

I'm running the classification method Bagging Tree (Bootstrap Aggregation) and compare this misclassification error rate with one from one single tree.
It's strange to me because the function estim.pred returns a matrix of factors that map to "pos" and "neg", but res.boot$t returns a matrix of integers taking on the values of 1 or 2, where as estim.pred is the statistic of res.boot$t.
Could you please explain the reason for this phenomenon?
library(rpart)
library(boot)
library(mlbench)
data(PimaIndiansDiabetes)
n <- 768
ntrain <- 468
ntest <- 300
B <- 100
M <- 100
train.error <- vector(length = M)
test.error <- vector(length = M)
bagging.error <- vector(length = M)
estim.pred <- function(a.sample, vector.of.indices)
{
current.train <- a.sample[vector.of.indices, ]
current.fitted.model <- rpart(diabetes ~ ., data = current.train, method = "class")
predict(current.fitted.model, test.set, type = "class")
}
fitted.tree <- rpart(diabetes ~ ., data = train.set, method = "class")
pred.train <- predict(fitted.tree, train.set, type = "class")
res.boot = boot(train.set, estim.pred, B)
head(pred.train)
head(res.boot$t)

Here is #Roland comment. I post it here to remove my question from unanswered list.
res.boot$t is a matrix. A matrix cannot contain a factor variable. Thus, the matrix contains the underlying integer values. Transpose the matrix, turn it into a data.frame and turn the integers into factor variables with your levels.

What's wrong with my nested for loop in R?

fname = file.choose()
two = read.csv(fname.header=T)
rec = two$Receipt
del = two$Delivery
date = two$Date
net = rec-del
yrec = matrix(rec,nrow=365,ncol=4,byrow=F)
ydel = matrix(del,nrow=365,ncol=4,byrow=F)
ynet = matrix(net,nrow=365,ncol=4,byrow=F)
yrecsum = 0
yrecavg = 0
for(i in 1:4)
{
for(j in 1:365)
{
yrecsum[i] = yrecsum[i]+yrec[j,i]
}
yrecavg[i] = yrecsum[i]/365
}
So what I have are three matrices of the same size with days in integers (from 1 to 365) on the rows and years integers (from 1 to 4) on the columns. Each matrix is filled in with the data that I'm working with.
I'm trying to find the average of each column for all three matrices and I would like to put those averages in a vector for each matrix.
I've looked around and found some information about the zoo library and chron library and such but I can't get those to work.

lapply(list(yrec, ydel, ynet), colMeans)
[[1]]
[1] 732.9370 731.9836 705.3808 751.6986
[[2]]
[1] 704.7178 714.2877 735.4822 767.5123
[[3]]
[1] 749.1041 715.4164 711.1425 746.3370
#Data
yrec <- matrix(sample(365*4), ncol=4)
ydel <- matrix(sample(365*4), ncol=4)
ynet <- matrix(sample(365*4), ncol=4)

this should get you started (even though I would convert the matrices to data.frames):
#some sample data
m <- matrix(sample(10000, 365*4),365,4)
# get the mean of all the columns of your matrix
colMeans(m)
if you have 3 matrices and you want to combine the results I would do:
# some sample data:
m1 <- matrix(sample(10000, 365*4),365,4)
m2 <- matrix(sample(10000, 365*4),365,4)
m3 <- matrix(sample(10000, 365*4),365,4)
do.call("cbind", lapply(list(m1,m2,m3), colMeans))

Resources