Scribd
Upload
30 views

Tutorial On Loops and Functions: September 28, 2007

This tutorial provides examples of using loops in data analysis problems. It demonstrates how to: 1) Generate random sample data in vectors and store them in a list using a for loop. 2) Perform all possible t-tests on the vector pairs using nested for loops to iterate through each pair. Results are stored in a data frame. 3) Write a function to take the exponential of a value multiple times using a for loop, demonstrating how the results grow very large with more iterations.

Uploaded by

ErdiApatay
Copyright
© © All Rights Reserved
Available Formats
Download as PDF, TXT or read online on Scribd
30 views

Tutorial On Loops and Functions: September 28, 2007

This tutorial provides examples of using loops in data analysis problems. It demonstrates how to: 1) Generate random sample data in vectors and store them in a list using a for loop. 2) Perform all possible t-tests on the vector pairs using nested for loops to iterate through each pair. Results are stored in a data frame. 3) Write a function to take the exponential of a value multiple times using a for loop, demonstrating how the results grow very large with more iterations.

Uploaded by

ErdiApatay
Copyright
© © All Rights Reserved
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 3

Tutorial on Loops and Functions

September 28, 2007

This tutorial gives a few examples of typical uses of loops in simple data analysis
problems.
Problem 1. In the first example, suppose we are given numerous vectors, say 10,
and asked to perform a t test on all possible pairs. As output, produce a table of
illustrative test components, sorted by p-value.
In practice we’d be given the vectors; here I’ll create some sample data. The 10
vectors will be stored in a list. First create an empty list. Then execute a loop that
generates a vector of length 20, randomly from normal distributions with varying
means. We start from a mean of −2 and increase it by 0.5 at each step. We’ll leave
the standard deviation at 1 (the default) in each sample.

> X <- vector(mode = "list", length = 10)


> m <- -2
> for (i in 1:10) {
+ X[[i]] <- rnorm(20, mean = m)
+ m <- m + 0.5
+ }
> names(X) <- paste("x", 1:10, sep = "")

We’re given 10 vectors, x1 , . . . , x10 and asked to perform a t test on each pair. If
we execute a t test on x1 , x2 , there is no need to repeat the test on x2 , x1 . Order only
matters in how certain numbers are reported (in a two-sided test). The following
array displays the pairs on which a t test should be run.
x1 x2 x1 x3 x1 x4 ··· x1 x10
x2 x3 x2 x4 ··· x2 x10
.. ..
. ··· .
x8 x9 x8 x10
x9 x10

That is, for each i, 1 ≤ i ≤ 9, we execute a t test on xi x j , for every j, i + 1 ≤ j ≤ 10.


We will need a loop inside a loop to run all these tests. Each iteration of the double

1
loop performs one t test. We’ll want to store selected components of the test object
as rows in a data frame. To keep track of the variables in that particular t test we
need to associate it with a descriptive name. As the row names of the data frame
we use the name “1-2” for the t test with x1 , x2 , and similarly for other variables. As
useful components of the t-test we select statistic, p.value, estimate.
To store to t test results we create data frame with 45 entries (the total number of
tests) having 0’s as the entries and the characters “1” to “45” as the rownames. The
counter l keeps track of which of the 45 tests we are running and identifies the row
in which data should be stored.

> testDat <- data.frame(Statistic = numeric(45), P.value = numeric(45),


+ Estimate = numeric(45))
> rownames(testDat) <- as.character(1:45)
> l <- 1
> for (i in 1:9) {
+ for (j in (i + 1):10) {
+ nm <- paste(i, j, sep = "-")
+ tst <- t.test(X[[i]], X[[j]], var.equal = TRUE)
+ rownames(testDat)[l] <- nm
+ testDat[l, 1:3] <- c(tst$stat, tst$p.val, tst$est)
+ l <- l + 1
+ }
+ }

Now sort the rows of testDat by p-value.

> testDat1 <- testDat[order(testDat$P.value), ]

The first 10 rows and the last 10 rows of the table are reported on the following
page.

Problem 2. Write a function in two variables, x and n, that successively takes the
exponential of x n times. NOTE: This function gets very large as n increases.

> iterExp <- function(x, n) {


+ y <- x
+ for (i in 1:n) {
+ y <- exp(y)
+ }
+ y
+ }

Samples:

2
Statistic P.value Estimate
1-10 −15.40 6.33e−18 −1.74
4-10 −14.00 1.41e−16 −0.89
1-9 −12.85 2.09e−15 −1.74
2-10 −12.79 2.44e−15 −0.96
3-10 −11.72 3.44e−14 −0.99
4-9 −11.13 1.60e−13 −0.89
1-8 −10.61 6.50e−13 −1.74
2-9 −10.43 1.04e−12 −0.96
3-9 −9.73 7.22e−12 −0.99
1-7 −9.27 2.69e−11 −1.74

Table 1: First 10 Rows

Statistic P.value Estimate


1-3 −2.55 1.48e−02 −1.74
3-5 −2.51 1.65e−02 −0.99
7-8 −2.35 2.38e−02 0.94
8-10 −2.24 3.11e−02 1.75
6-7 −1.94 6.00e−02 0.36
5-6 −1.92 6.20e−02 −0.21
8-9 −1.16 2.53e−01 1.75
9-10 −1.07 2.90e−01 2.16
3-4 −0.41 6.83e−01 −0.99
2-4 −0.33 7.42e−01 −0.96
2-3 0.11 9.15e−01 −0.96

Table 2: Last 10 Rows

> iterExp(5, 2)

[1] 2.851124e+64

> iterExp(2, 5)

[1] Inf

You might also like