Statistics for Data Science - 2

Graded assignment Solution- Sept 2024

Week 7

1. Suppose X1 , X2 , . . . , Xn are n iid random variables with mean µ and variance σ 2 = 16.
Using Chebyshev’s inequality, find the minimum value of n such that

P (| X − µ |< 1) > 0.90.

Answer: 160
Solution:
σ2
E[X̄] = µ, Var[X̄] =
n
Note that X̄ = X1 + X2 + . . . + Xn
By Chebyshev’s inequality, we have

σ2
P X −µ <δ ≥1− 2
nδ
And given

P (| X − µ |< 1) > 0.90

σ2
⇒ 1 − 2 > 0.90
nδ
16
⇒1− > 0.90
n
16
⇒ 1 − 0.90 >
n
16
⇒ 0.10 >
n
16
⇒n>
0.10
⇒ n > 160

Hence, the minimum of n should be 160.

2. Consider a sample 1, 0, 1, 0, 1, 1, 0, 1, 1, 1 from the Bernoulli(0.6) distribution.

(a) Compute the empirical distribution of the sample.
A. p(0) = 0.4, p(1) = 0.6
B. p(0) = 0.3, p(1) = 0.7
C. p(0) = 0.5, p(1) = 0.5
Course: Statistics for Data Science - II Page 2 of 11

D. p(0) = 0.7, p(1) = 0.3

Answer: B

Solution :
Since, the empirical distribution is the discrete distribution with PMF

#(Xi = t)
p(t) =
n
#(Xi = t) denotes the number of times t occurs in the samples.
Therefore,
3
p(0) = = 0.3
10
And
7
p(1) = = 0.7
10
Hence, option (B) is correct.

(b) Compute the sample mean. Enter the answer correct to one decimal place.
Answer: 0.7
Solution:
X 1 + X2 + . . . + X 1 0
X̄ =
n
1+0+1+0+1+1+0+1+1+1
=
10
7
=
10
= 0.7

3. Let X1 , X2 , X3 are three independent and identically distributed random variables with
mean µ and variance σ 2 . Given below are 3 different formulations of sample mean.
(Observe that E[A] = E[B] = E[C]).

X1 + X 2 + X3
A=
3

B = 0.1X1 + 0.3X2 + 0.6X3

C = 0.2X1 + 0.3X2 + 0.5X3

Course: Statistics for Data Science - II Page 3 of 11

Choose the correct option from the following:

(a) Var(A) = Var(B) = Var(C)
(b) Var(A) ≥ Var(B) ≥ Var(C)
(c) Var(A) ≤ Var(B) ≤ Var(C)
(d) Var(A) ≤ Var(C) ≤ Var(B)

Solution:
Let X1 , X2 , X3 ∼ i.i.d. X, where E[X] = µ, Var(X) = σ 2

 
X1 + X2 + X3
Var(A) = Var
3
1
= (Var[X1 ] + Var[X2 ] + Var[X3 ])
9
1 σ2
= (3σ 2 ) =
9 3

Var(B) = Var(0.1X1 + 0.3X2 + 0.6X3 )

= 0.01Var[X1 ] + 0.09Var[X2 ] + 0.36Var[X3 ]
= 0.46σ 2

Var(C) = Var(0.2X1 + 0.3X2 + 0.5X3 )

= 0.04Var[X1 ] + 0.09Var[X2 ] + 0.25Var[X3 ]
= 0.38σ 2

Therefore, Var(B) ≥ Var(C) ≥ Var(A).

4. A random sample of size 25 is collected from a normal population with a mean of 50

and a standard deviation of 5. Find the variance of the sample mean.
Solution:
We know that the variance of the sample mean X is given by

σ2
Var[X] =
n
52
=
25
=1
Course: Statistics for Data Science - II Page 4 of 11

X
5. A fair die is rolled 100 times. Let denote the number of times six is obtained. Find
100
X 1
a bound for the probability that differs from by less than 0.1 using the weak law
100 6
of large numbers.

5
(a) at least
36
31
(b) at least
36
5
(c) at most
36
31
(d) at most
36
Solution:
X denotes the number of times six is obtained on rolling a fair die 100 times. Let
X1 , X2 , . . . , X100 be 100 i.i.d. samples such that
(
1 if six appears on rolling a fair die
Xi =
0 otherwise

1 5
E[Xi ] = µ = and Var(Xi ) = σ 2 =
6 36
Notice that X = X1 + X2 + X3 + · · · + X100 .
 
X 1
To find: Bound on P − < 0.1 . By the weak law of large numbers, we have
100 6
σ2
P (|X − µ| < δ) ≥ 1 − 2
nδ

  5
X 1 36
⇒P − < 0.1 ≥ 1 −
100 6 100 × 0.01
 
X 1 5 31
⇒P − < 0.1 ≥ 1 − =
100 6 36 36

6. Let X1 , X2 , . . . , X5 be i.i.d. samples whose distribution has a mean of 20 and variance

of 4. Suppose the sample variance is defined as

(X1 − X)2 + · · · + (X5 − X)2

S2 =
5
X1 +X2 +···+X5
where X = 5
. Find the expected value of S 2 .
Course: Statistics for Data Science - II Page 5 of 11

Solution:

σ2 4
E[X] = µ = 20 and Var[X] = = = 0.8.
n 5
" n
#
1 X
E[S 2 ] = E (Xi − X)2
n i=1
" n #
1 X
= E (Xi − X)2
n
" i=1
n
#
1 X 2
= E (Xi2 + X − 2Xi X)
n
" i=1
n
#
1 X 2
= E Xi2 + nX − 2nXX
n
" i=1
n
#
1 X 2 2
= E Xi2 + nX − nX
n
" i=1
n
#
1 X
= E Xi2
n
" ni=1 #
1 X 2
= E[Xi2 ] − nE[X ]
n i=1
" n  2 #
1 X 2 σ
= (σ + µ2 ) − n + µ2
n i=1 n
1
(nσ 2 + nµ2 ) − (σ 2 + nµ2 )

=
n
(n − 1)σ 2
=
n

4
Here, n = 5, therefore, E[S 2 ] = × 4 = 3.2.
5
7. Suppose Xi ∼ Normal 0, i12 , where i = 1, 2, . . . , 9 and X1 , X2 , . . . , X9 are independent

of each other. Let Y be a random variable defined as Y = 9i=1 iXi . Find the variance
P
of Y .
Solution
Course: Statistics for Data Science - II Page 6 of 11

9
!
X
Var(Y ) = Var iXi
i=1
= Var(X1 + 2X2 + 3X3 + · · · + 9X9 )
= Var(X1 ) + Var(2X2 ) + · · · + Var(9X9 )
= Var(X1 ) + 4Var(X2 ) + · · · + 81Var(X9 )
   
1 1 1
= 2 + 4 2 + · · · + 81 2
1 2 9
=9

8. A random sample of size 50 is collected from a population P , where P ∼ Uniform[0, 12].

Find a lower bound on the probability that the sample mean will be at most 3 units
away from the actual mean using the weak law of large numbers.

Solution:

P ∼ Uniform[0, 12]
0 + 12 (12 − 0)2 144
E[P ] = µ = = 6, Var(P ) = σ 2 = = = 12
2 12 12

By weak law of large numbers, we have

σ2
P (|X − µ| < δ) ≥ 1 −
nδ 2

12 73
P |X − µ| < 3 ≥ 1 − = ≈ 0.9733
50 × 9 75

9. Suppose a random sample is used to estimate the proportion of voters in a city. If the
sample proportion is roughly 0.45, what sample size is necessary so that the standard
deviation of the sample proportion is 0.02?
Solution
Let the random variable X represent that the selected candidate is a voter.
Let Xi be defined as
(
1, if the selected candidate is a voter
Xi =
0, otherwise

Define an event A as A : X = 1.
It is given that P (A) = 0.45.
P (A)(1 − P (A))
We know that Var(S(A)) = .
n
Course: Statistics for Data Science - II Page 7 of 11

r
p(1 − p)
= 0.02
r n
(0.45)(0.55)
= 0.02 =⇒ n = 618.75 ≈ 619
n

10. The average life (in years) of an electronic watch follows an exponential distribution with
1
parameter . Find the lower bound on the probability that the mean life of a random
2
sample of 50 such watches falls between 1 and 3 years. Enter your answer correct to two
decimals.
Hint: Use the weak law of large numbers.
Solution
Let the random variable X represent the life of an electronic watch.
It is given that X ∼ Exp(1/2) and 50 such samples are taken.
E[X] = µ = 2, Var(X) = σ 2 = 4
To find: a lower bound on P (1 < X < 3).
By the weak law of large numbers, we have

σ2
P (|X − µ| < δ) ≥ 1 −
nδ 2

4
P |X − 2| < 1 ≥ 1 −
50 × 1
23
=
25
= 0.92

11. A university evaluates the final scores of students based on coursework and a final project.
The variance of the coursework scores is 15, and the variance of the final project scores
is 30. Coursework contributes 70% to the final evaluation, while the project contributes
30%. Assuming the scores of coursework and project are uncorrelated, what is the
variance of the final evaluation scores? Enter the answer correct to two decimal places.
Answer : 10.05 ; Range : 10.02 to 10.08
Solution
To compute the variance of the final evaluation scores, we use the formula for the variance
of a linear combination of uncorrelated random variables:

Var(aX + bY ) = a2 Var(X) + b2 Var(Y ),

where X and Y are independent random variables, and a and b are the weights of X
and Y , respectively.
Course: Statistics for Data Science - II Page 8 of 11

In this case:

a = 0.7 (weight of coursework),

b = 0.3 (weight of the project),
Var(X) = 15 (variance of coursework scores),
Var(Y ) = 30 (variance of project scores).

Substitute the values into the formula:

Var(Final Evaluation) = (0.7)2 · 15 + (0.3)2 · 30.

Var(Final Evaluation) = 0.49 · 15 + 0.09 · 30.

Var(Final Evaluation) = 7.35 + 2.7 = 10.05.

Therefore, the variance of the final evaluation scores is 10.05.

12. In a large population of students, an unknown proportion p prefers Chips over Cookies.
A survey was conducted among 200 students and 120 preferred Chips. Assume the 200
samples are i.i.d. Bernoulli (p).
Based on the given information, answer the following questions:
[(i).]What is the sample mean? Enter the answer correct to one decimal place.
Answer : 0.6 To determine the sample mean, we use the formula for the sample
mean:
Number of successes
X̄ =
Total number of samples
Given:

Number of successes = 120 (students who preferred Chips),

Total number of samples = 200 (total students surveyed).

Substituting the values into the formula:

120
X̄ = .
200
Simplify the fraction:
X̄ = 0.6.
Thus, the sample mean is:
0.6 .
What is the variance of the sample mean? Assume p is equal to the sample
mean.[(a)]
2.
1. (a) 0.24
Course: Statistics for Data Science - II Page 9 of 11

(b) 0.36
(c) 0.0018
(d) 0.0012
Answer : d
The variance of the sample mean is given by the formula:

p(1 − p)
Var(X̄) = ,
n
where:

p = proportion of successes (equal to the sample mean),

n = total number of samples.

From the given data:

p = 0.6 (calculated sample mean),

n = 200 (total number of students surveyed).

Substituting the values into the formula:

0.6(1 − 0.6)
Var(X̄) = .
200
Simplify the expression:
0.6 · 0.4
Var(X̄) = .
200

0.24
Var(X̄) = .
200

Var(X̄) = 0.0012.
Thus, the variance of the sample mean is:

0.0012 .

3. If 30 more students join the survey and they all prefer Cookies, what will be the
new sample mean and variance of the sample mean?
[(a)]new sample mean = 0.24, new variance = 0.0012 new sample mean = 0.6,
new variance = 0.0012 new sample mean = 0.6522, new variance = 0.0009 new
sample mean = 0.5217, new variance = 0.00108
Answer : d
Course: Statistics for Data Science - II Page 10 of 11

New Sample Mean

The formula for the sample mean is:
Number of successes
X̄ = .
Total number of samples

From the given data:

Original number of successes = 120,

Original total number of samples = 200.

If 30 more students join and they all prefer Cookies, the updated values are:

New number of successes = 120 (unchanged, as no additional students prefer Chips),

New total number of samples = 200 + 30 = 230.

The new sample mean is:

120
X̄ = .
230
Simplify the fraction:

X̄ = 0.5217 (rounded to 4 decimal places).

Thus, the new sample mean is:

0.52174 .

New Variance of the Sample Mean

The formula for the variance of the sample mean is:
p(1 − p)
Var(X̄) = ,
n
where:

p = new sample mean,

n = new total number of samples.

Substituting the values:

p = 0.52174,
n = 230.

Substitute these into the formula:

0.52174(1 − 0.52174)
Var(X̄) = .
230
Course: Statistics for Data Science - II Page 11 of 11

Simplify the calculation:

0.52174 · 0.47826
Var(X̄) = .
230

0.249998
Var(X̄) = .
230

Var(X̄) = 0.00108 (rounded to 5 decimal places).

Thus, the new variance of the sample mean is:

0.00108 .