1. Generate a large dataset ( at least 1000 observations) µ with a known mean, µ between 20 and 40 and variance s = 9 using rnorm(). identify it as D. Use the functions mean() and var() to get the mean and vaiance of the sample.
2. Use t.test(D, mu=0) to check if the mean of your realizations, ¯x is signi?cantly
different from zero, from µ1, and from (µ0:1).
3. produce a histogram for D with vertical lines at the 0.25
th
and the 97.5
th
quantiles.
4. generate 10 samples of 20, S1 .....S10 with random means using S1<- rnorm(20,runif(1,min=0,max
= 80)), then calculate means and variances for each.
5. Use t.test(D, S1) to check if the sample means for S1 .....S10 are likely to have come
from the same population as the large distribution D. This Stat Trek webpage is a
good source. You should be using 2-tailed tests.
6. Explain the difference between a one-tailed test and a two-tailed test.
7. Create two variables, x and y, where y is a deterministic linear function of x. Show
the formula you used.
8. Generate a plot of y against x.
9. Calculate the correlation between y and x.covariance of y and x.
10. Create a new variable, yRan, that is like y except that it contains some random noise.
11. Calculate the correlation between yRan and x and the covariance of yRan and x.
12. Add more noise to yRan. Calculate the covariance of yRan and x and the Pearsonian correlation coef?cient. State how adding more noise affects the correlation
coef?cient.
13. Conduct the following experiment. Start with a deterministic function as in part 8
with a positive slope. Add some noise. Now systematically reduce the slope from
positive to negative, taking care to make quite a few observations near zero. Keep
the noise parameters constant. Make a table showing the slope vs the correlation
coef?cient.
14. Plot the slope vs the correlation coef?cient from the previous question and say what
you learn from the exercise.