Assignment

Probability & Statistics

Mridul Sarmah

Indian Institute of Management, Udaipur

Roll No: 2421007

01-10-2024
 6.(b) Solution
In [ ]: import numpy as np
import matplotlib.pyplot as plt

# Step 1: Generate 1000 random samples using the inverse CDF method
n = 1000
u = np.random.uniform(0, 1, n)
x = np.sqrt(u) # Apply the inverse CDF: X = sqrt(U)

# Step 2: Plot the histogram of the generated random variable

plt.figure(figsize=(8, 6))
plt.hist(x, bins=30, density=True, alpha=0.6, color='b', label="Empirical Densit

# Step 3: Plot the theoretical PDF

# The PDF is derived by differentiating the CDF F(x) = x^2 for 0 ≤ x ≤ 1.
# PDF: f(x) = 2x for 0 ≤ x ≤ 1
x_vals = np.linspace(0, 1, 100)
pdf_vals = 2 * x_vals # Theoretical PDF

plt.plot(x_vals, pdf_vals, 'r-', lw=2, label="Theoretical PDF")

# Step 4: Add labels and legends

plt.xlabel('x')
plt.ylabel('Density')
plt.title('Empirical Density vs Theoretical PDF')
plt.legend()

# Show the plot

plt.show()
 8.(b) Solution
In [ ]: import numpy as np
import matplotlib.pyplot as plt

# Parameters
P = 0.48 # True population proportion
n = 1500 # Sample size
num_samples = 100 # Number of samples
confidence_level = 0.98 # 98% confidence level
Z = 2.326 # Z-value for 98% confidence level

# Function to calculate a confidence interval for each sample

def confidence_interval(sample_proportion, Z, n):
se = np.sqrt(sample_proportion * (1 - sample_proportion) / n)
return sample_proportion - Z * se, sample_proportion + Z * se

# Simulating 100 samples and calculating confidence intervals

intervals = []
proportions = []
contains_P = []
for _ in range(num_samples):
# Generate a random sample of size 1500 from a binomial distribution
sample = np.random.binomial(n, P) / n
proportions.append(sample)
# Calculate the confidence interval for this sample
ci_lower, ci_upper = confidence_interval(sample, Z, n)
intervals.append((ci_lower, ci_upper))
# Check if the true population proportion P is within the interval
contains_P.append(ci_lower <= P <= ci_upper)

# Plotting the confidence intervals and sample proportions with flipped axes
plt.figure(figsize=(10, 8))
for i, (ci_lower, ci_upper) in enumerate(intervals):
if contains_P[i]:
plt.plot([ci_lower, ci_upper], [i, i], color='blue') # Interval contain
else:
plt.plot([ci_lower, ci_upper], [i, i], color='red') # Interval doesn't

# Plot the sample proportion estimates as markers (on the y-axis)

plt.scatter(proportions, range(num_samples), color='black', zorder=5, label='Sam

# Plot a vertical line for the true population proportion P

plt.axvline(x=P, color='green', linestyle='--', label='True P (0.48)')

# Add labels and legend

plt.xlabel('Proportion')
plt.ylabel('Sample Index')
plt.title(f'98% Confidence Intervals and Sample Proportions for {num_samples} Sa
plt.legend()

# Show the plot

plt.show()
 9.(a) Solution
In [ ]: import numpy as np
import matplotlib.pyplot as plt
from scipy.stats import norm

# Parameters
n_samples = 25 # Number of independent samples in each trial
num_trials = 1000 # Number of trials
a, b = 0, 1 # Uniform distribution range [0, 1]

# Step 1: Simulate 1000 sums of 25 independent uniform random variables

sums = [np.sum(np.random.uniform(a, b, n_samples)) for _ in range(num_trials)]

# Step 2: Compute mean and standard deviation of the sums

mean_s25 = np.mean(sums)
std_s25 = np.std(sums)

# Step 3: Plot the histogram of the sums

plt.figure(figsize=(10, 6))
plt.hist(sums, bins=30, density=True, alpha=0.6, color='b', label="Histogram of

# Step 4: Plot the normal distribution with same mean and std
x = np.linspace(min(sums), max(sums), 100)
plt.plot(x, norm.pdf(x, mean_s25, std_s25), 'r', lw=2, label=f'Normal Density\nM

# Step 5: Add labels and legend

plt.title('Sums of 25 Uniform Random Variables repeated 1000 times\nAnd\n Corres
plt.xlabel('Sum of 25 Uniform Variables')
plt.ylabel('Density')
plt.legend()

# Show the plot

plt.show()
 9.(b) Solution
In [ ]: import numpy as np
import matplotlib.pyplot as plt
from scipy.stats import norm

# Parameters
n_samples = 25 # Number of independent samples in each trial
num_trials = 1000 # Number of trials

# Step 1: Inverse transform sampling for f(x) = 2x

def sample_from_fx_2x(size):
u = np.random.uniform(0, 1, size) # Generate uniform random numbers
return np.sqrt(u) # Apply the inverse CDF transformation

# Step 2: Simulate 1000 sums of 25 independent random variables

sums = [np.sum(sample_from_fx_2x(n_samples)) for _ in range(num_trials)]

# Step 3: Compute mean and standard deviation of the sums

mean_s25 = np.mean(sums)
std_s25 = np.std(sums)

# Step 4: Plot the histogram of the sums

plt.figure(figsize=(10, 6))
plt.hist(sums, bins=30, density=True, alpha=0.6, color='b', label="Histogram of

# Step 5: Plot the normal distribution with the same mean and standard deviation
x = np.linspace(min(sums), max(sums), 100)
plt.plot(x, norm.pdf(x, mean_s25, std_s25), 'r', lw=2, label=f'Normal Density\nM

# Step 6: Add labels and legend

plt.title('Sums of 25 Random Variables (f(x)=2x) repeated 1000 times\n And\n Cor
plt.xlabel('Sum of 25 Variables')
plt.ylabel('Density')
plt.legend()

# Show the plot

plt.show()
 9.(c) Solution
In [ ]: import numpy as np
import matplotlib.pyplot as plt
from scipy.stats import norm

# Parameters
n_samples = 25 # Number of independent samples in each trial
num_trials = 1000 # Number of trials

# Step 1: Inverse transform sampling for f(x) = 3x^2

def sample_from_fx_3x2(size):
u = np.random.uniform(0, 1, size) # Generate uniform random numbers
return u**(1/3) # Apply the inverse CDF transformation

# Step 2: Simulate 1000 sums of 25 independent random variables

sums = [np.sum(sample_from_fx_3x2(n_samples)) for _ in range(num_trials)]

# Step 3: Compute mean and standard deviation of the sums

mean_s25 = np.mean(sums)
std_s25 = np.std(sums)

# Step 4: Plot the histogram of the sums

plt.figure(figsize=(10, 6))
plt.hist(sums, bins=30, density=True, alpha=0.6, color='b', label="Histogram of

# Step 5: Plot the normal distribution with the same mean and standard deviation
x = np.linspace(min(sums), max(sums), 100)
plt.plot(x, norm.pdf(x, mean_s25, std_s25), 'r', lw=2, label=f'Normal Density\nM

# Step 6: Add labels and legend

plt.title('Sums of 25 Random Variables (f(x)=3x^2) repeated 1000 times\nAnd\n Co
plt.xlabel('Sum of 25 Variables')
plt.ylabel('Density')
plt.legend()

# Show the plot

plt.show()
 9.(d) Solution
In [ ]: import numpy as np
import matplotlib.pyplot as plt
from scipy.stats import norm

# Parameters
n_samples = 25 # Number of independent samples in each trial
num_trials = 1000 # Number of trials

# Step 1: Inverse transform sampling for f(x) = 4|x - 1/2|

def sample_from_fx_4_abs_x_minus_1_2(size):
u = np.random.uniform(0, 1, size) # Generate uniform random numbers
# Apply inverse CDF transformation
samples = np.where(u <= 0.5, 1-np.sqrt(1-u), (1 - np.sqrt(2*u)) / 2)
return samples

# Step 2: Simulate 1000 sums of 25 independent random variables

sums = [np.sum(sample_from_fx_4_abs_x_minus_1_2(n_samples)) for _ in range(num_t

# Step 3: Compute mean and standard deviation of the sums

mean_s25 = np.mean(sums)
std_s25 = np.std(sums)

# Step 4: Plot the histogram of the sums

plt.figure(figsize=(10, 6))
plt.hist(sums, bins=30, density=True, alpha=0.6, color='b', label="Histogram of

# Step 5: Plot the normal distribution with the same mean and standard deviation
x = np.linspace(min(sums), max(sums), 100)
plt.plot(x, norm.pdf(x, mean_s25, std_s25), 'r', lw=2, label=f'Normal Density\nM

# Step 6: Add labels and legend

plt.title('Sums of 25 Random Variables (f(x) = 4|x - 1/2|) repeated 1000 times\n
plt.xlabel('Sum of 25 Variables')
plt.ylabel('Density')
plt.legend()

# Show the plot

plt.show()
 6.
The e
ypoesia

Po.8

Ptm (outicod vale)

As the Lineny
Populatiou hottoy

inomia Dici kniou,

he beeple to bollotd
fe
ino i ol henee the ean

and Vauane Wold de

Ho

Since Aize
C>30)

M.

(-e./

np(0-P)
 M- 100Xo.z 2o.01
NJooxo.1X02

P2< 4
tu 2-Atehe
tale

m- 80 -2.93
4
80-9.32
W
= 70. 63

71-100X0.7

V]o0 x 0.9 2x0.22)

= P(22 -l-69)

=1- (I- (22149))

: P(241-69 )
2- uble)
= 0.9 545
 7)- J0oX 0.7C
r (?2 NJoox0.7s025

(?2 -0.92)

0.81

71- )00X 0.72

P(? 160x0.9Lx 0.27

= P(2 -0.1)
P(2< O. 22)

D.527

(Y) P 0.70

Nj00x 0.90o.3
= P(22 0.22)

) 1 -.ET1
0 129
(v) = 0.48
7- J00X 0.61
P(2 2 )

|-r(2 20.64)

P( 22 ?|-)0 x0.45
Nlo0 xo-6sX0.3S

|- 0.39%2

b.103

(i) P= 0.63
-100x o. 63
Noxo.63 20.27

)-1(22|.·36)
| 0.95)S
,04 85
 - P(?2 2-24)
- )-P(2can2u)
o.9 B75
o.o125

ot the
Aee thot the pokabi ity
he ml hypo fheas
reases
Cie. Pe tr eoh) dese de
We decheuse
igibi canty dhen
ole e u
to 0.6. So, TyPe I! eoh

we move

menKes

Cen Ahy that fhe test in vnoat ponbul.

Hence, ve