Scribd
Upload
12 views3 pages

MTH211A Problem Set-2

Uploaded by

Tisha Ghosh
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
12 views3 pages

MTH211A Problem Set-2

Uploaded by

Tisha Ghosh
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 3

MTH211A: Theory of Statistics

Problem set 2

1. Let there be two sets of samples on the variable x. The j-th sample is of size nj ,
mean x̄j and median Mj , j = 1, 2. Let the mean and median of the pooled sample
of size n1 + n2 be x̄ and M , respectively. Show that

(a) min {x̄1 , x̄2 } ≤ x̄ ≤ max {x̄1 , x̄2 },

(b) min {M1 , M2 } ≤ M ≤ max {M1 , M2 }.

2. Let x be a variable, g : R → R be a function and y = g(x). Suppose x1 , . . . , xn are


n observations on x, and yi = g(xi ), for each i. Show that

(a) ȳ = g(x̄) if g is linear.

(b) ȳ ≥ g(x̄) if g is convex, and ȳ ≤ g(x̄) if g is concave.

(c) ỹme = g(x̃me ) if g is monotone.

(d) ỹmo = g(x̃mo ) if g is one-to-one.

3. Suppose x1 , . . . , xn are n observations on a variable x, and yi = a + bxi , where a ∈ R


and b ∈ R \ {0}. Let, for some positive integer r, {mr (x), m′r (x)} be the r-th central
and raw moments of x, and {mr (y), m′r (y)} be those of y.

(a) Express m′r (y) in terms of m′s (x), where s = 1, . . . , r.

(b) Express mr (y) in terms of mr (x).

(c) Hence or otherwise find the relation between MDx̄ (x) and MDȳ (y); SD(x) and
SD(y); g1 (x) and g1 (y).

1
(d) Let x̃q and ỹq be the q-th qunatile of x and y, respectively. Find the relation
between x̃q and ỹq .

(e) Hence or otherwise, find the relation between QD(x) and QD(y); SK3 (x) and
SK3 (y).

4. Let a and b be two vectors in Rn . Show that


Pn
(a) [Cauchy-Schwarz inequality] a · b ≤ ∥a∥2 × ∥b∥2 , where a · b = i=1 ai bi and
pPn
2
∥a∥2 = i=1 ai .

(b) Suppose n paired observations, {(xi , yi ), i = 1, . . . , n} on two variables x and


y are available. Let c and d be two real numbers. Show that var(cx + dy) =
c2 var(x) + d2 var(y) + 2cd cov(x, y), where cov(x, y) = n−1 ni=1 (xi − x̄)(yi − ȳ).
P

(c) Using part (a) or otherwise, show that SD(x + y) ≤ SD(x) + SD(y).
Pn
(d) Define ∥a∥1 = i=1 |ai |. Using part (a) or otherwise, show that ∥a∥1 ≤
√
n∥a∥2 .

(e) Using part (d), show that MDx̄ (x) ≤ SD(x).

5. Suppose x1 , . . . , xn are n observations of a variable x.

(a) Show that SD is the least root-mean-square deviation of x.

(b) Let there be k samples on the variable x. The sample size, mean and variance
of the j-th sample are nj , x̄j and s2j , j = 1, . . . , n. Express the variance of the
combined sample in terms of nj , x̄j and s2j , j = 1, . . . , n.
Pn Pn
(c) Let ∆ be the Gini’s mean-squared difference measure defined as ∆ = n−2 i=1 j=1 (xi −

xj )2 . Show that ∆ = 2var(x).

6. Suppose x1 , . . . , xn are n observations of a variable x, and R and SK2 are the range
and Pearson’s measure of skewness of second type, respectively. Show that

(a) (2n)−1 R2 ≤ var(x) ≤ 4−1 R2 .

(b) −3 ≤ SK2 ≤ 3.

2
7. Suppose the central moments of a variable x for first r integers, ms , s = 1, . . . , r,
are calculated. However, you want to calculate the first r moments of x about the
median x̃me . If only the difference between mean and median, (x̄ − x̃me ), is known
along with the central moments, will you be able to calculate the first r moments
of x about x̃me ?

8. Suppose a variable x takes positive values only, and the deviations (xi − x̄) is small
compared to x̄, such that |xi − x̄|/x̄ ≤ cn−1/2 , for some constant c and each i. Then
show that
lim x̄ = lim xg = lim xh .
n→∞ n→∞ n→∞

How do you interpret this result?

You might also like