9 Probability and Statistics

Author

Affiliation

Md Rasel Biswas

ISRT, University of Dhaka

1 Generating random data

• The function sample() is used to generate random values from a vector, and it has the following arguments:

  • x \(\rightarrow\) A vector of outcome you want to sample from
  • size \(\rightarrow\) The number of samples (observations) you want to draw
  • replace \(\rightarrow\) It can take either TRUE or FALSE
  • prob \(\rightarrow\) Specifies probability of selection of different elements of x

sample(x = 1:10, size = 4, replace = F)
#> [1] 1 2 3 4

---

Select 10 numbers from 0 to 100

sample(x = 0:100, size = 10, replace = F) # replace=FALSE
#>  [1] 52 61 82 41 78 66 49 57 12  7

sample(x = 0:100, size = 10, replace = T) # replace=TRUE
#>  [1] 78 56 75 59 66 70  9 32 81 72

---

• Select students’ grades randomly

sample(replace = TRUE, x = LETTERS[1:4], size = 10)
#>  [1] "A" "A" "C" "C" "B" "A" "A" "A" "C" "B"

• Tossing a fair coin 10 times

sample(replace = TRUE, x = c("H", "T"), size = 10)
#>  [1] "T" "H" "T" "T" "T" "H" "T" "T" "T" "T"

• Tossing a biased coin 10 times

sample(replace = TRUE, x = c("H", "T"), size = 10, prob = c(.7, .25))
#>  [1] "H" "H" "T" "H" "H" "T" "H" "H" "H" "H"

2 Use of initial seed in generating random numbers

Without seed:

# No seed
sample(1:10, 3)
#> [1] 1 2 5
# No seed
sample(1:10, 3)
#> [1] 4 5 7
# No seed
sample(1:10, 3)
#> [1] 4 1 7

With seed:

set.seed(100)
sample(1:10, 3)
#> [1] 10  7  6
set.seed(100)
sample(1:10, 3)
#> [1] 10  7  6
set.seed(100)
sample(1:10, 3)
#> [1] 10  7  6

3 Useful functions related to probability distributions:

• R has built in many functions for conveniently working with a large number of distributions.

• The quantities that are of main interest from any probability distribution are:

  • Probability density function (pdf for continuous variable) or probability mass function (pmf for discrete variable)
  • Cumulative distribution function (cdf)
  • Quantile function (inverse cdf)
  • Generating random sample from respective distributions.

---

Distribution	Density Function	Cumulative Distribution	Quantile	Random Variates
Normal	dnorm()	pnorm()	qnorm()	rnorm()
Poisson	dpois()	ppois()	qpois()	rpois()
Binomial	dbinom()	pbinom()	qbinom()	rbinom()
Uniform	dunif()	punif()	qunif()	runif()

• Such functions are available for other probability distributions, such as exponential, logistic, Chi-squared etc.

4 rbinom() and rnorm

• rbinom() is used to draw a sample from a binomial distribution

  • size \(\rightarrow\) number of Bernoulli trials

  • prob \(\rightarrow\) probability of success

  • n \(\rightarrow\) number of observations

• Draw a sample of size 8 from \(B(10, 0.75)\)

rbinom(size = 10, prob = .75, n = 8)
#> [1] 9 8 8 6 8 7 9 7

---

• rnorm() is used to draw a sample from a normal distribution

  • mean \(\rightarrow\) mean of the distribution \((\mu)\)

  • sd \(\rightarrow\) standard deviation of the distribution \((\sigma)\)

  • n \(\rightarrow\) number of observations

• Draw a sample of size 5 from \(N(10, 16)\)

rnorm(mean = 10, sd = 4, n = 5)
#> [1] 14.743527  8.970948 11.748854  8.539669 11.986696

5 pnorm()

• For \(X \sim N(50, 3^2)\), find \(P(45<X<55)\).

• \(P(a < X ≤ b) = F(b) − F(a)\)

pnorm(q = 55, mean = 50, sd = 3) - 
  pnorm(q = 45, mean = 50, sd = 3) 
#> [1] 0.9044193

6 dnorm()

• For \(X \sim Bin(10, 0.5)\), find \(P(X=5)\).

dbinom(x = 5, size = 10, prob = 0.5)
#> [1] 0.2460938

7 qnorm()

• Let \(Z\) follows a standard normal distribution. Then the 0.975−quantile is \(Z_{0.975} ≈ 1.96\). It means the probability of sampling a value less than or equal to 1.96 is 0.975 or \(97.5%\)

qnorm(p = 0.975, mean = 0, sd = 1)
#> [1] 1.959964