SEHH2311

FOUNDATIONS OF DATA SCIENCE

LECTURE 3
F, t and 𝝌𝝌𝟐𝟐 Distributions
 Topics
1. 𝜒𝜒 2 distribution
2. t distribution
3. F distribution

SEHH2311 Foundations of Data Science Page 2

 𝟐𝟐
𝝌𝝌 Distribution
• Suppose 𝑍𝑍1 , 𝑍𝑍2 , …, 𝑍𝑍𝑘𝑘 are iid 𝑁𝑁(0,1) random variables.
• If 𝑋𝑋 = 𝑍𝑍12 + 𝑍𝑍22 + ⋯ + 𝑍𝑍𝑘𝑘2 , then 𝑋𝑋 has a chi-square distribution with 𝑘𝑘 degrees
of freedom, denoted by 𝝌𝝌𝟐𝟐 (𝒌𝒌).
• pdf of 𝜒𝜒 2 random variable
1 𝑘𝑘
−1 −
𝑥𝑥
𝑓𝑓 𝑥𝑥 = 𝑥𝑥 2 𝑒𝑒 2 , 0 ≤ 𝑥𝑥 < ∞
𝑘𝑘
Γ 2 𝑘𝑘/2
2
∞
where Γ 𝑧𝑧 = ∫0 𝑢𝑢 𝑧𝑧−1 𝑒𝑒 −𝑢𝑢 𝑑𝑑𝑑𝑑 (Gamma Function)
• pdf of 𝜒𝜒 2 is skewed to the right.
• CDF of 𝜒𝜒 2 random variable
𝑥𝑥
𝐹𝐹 𝑥𝑥 = ∫0 𝑓𝑓 𝑢𝑢 𝑑𝑑𝑑𝑑 (This is somewhat complicated)
• 𝐸𝐸 𝑋𝑋 = 𝑘𝑘
• 𝑉𝑉𝑉𝑉𝑉𝑉 𝑋𝑋 = 2𝑘𝑘

SEHH2311 Foundations of Data Science Page 3

 𝟐𝟐
pdf of 𝝌𝝌 Distribution
pdf of 𝜒𝜒 2 with different degrees of freedom

https://en.wikipedia.org/wiki/Chi-squared_distribution

SEHH2311 Foundations of Data Science Page 4

 𝟐𝟐
CDF of 𝝌𝝌 Distribution
CDF of 𝜒𝜒 2 with different degrees of freedom

https://en.wikipedia.org/wiki/Chi-squared_distribution

SEHH2311 Foundations of Data Science Page 5

 𝟐𝟐
Using 𝝌𝝌 Distribution Table
𝜒𝜒 2 Distribution Table is typically used 𝜒𝜒 2 Distribution with 5 d.f.
when constructing confidence intervals
and testing hypotheses.
Example:
If 𝑋𝑋 is 𝜒𝜒 2 (5), then 𝐹𝐹 1.61 = 0.1.
or d.f.
2
𝜒𝜒0.9; 5 = 1.61
1 − 𝐹𝐹(𝑥𝑥)

F(x) 0.005 0.010 0.025 0.050 0.100 0.900 0.950 0.975 0.990 0.995
1-F(x) 0.995 0.990 0.975 0.950 0.900 0.100 0.050 0.025 0.010 0.005
d.f.
1 0.00 0.00 0.00 0.00 0.02 2.71 3.84 5.02 6.63 7.88
2 0.01 0.02 0.05 0.10 0.21 4.61 5.99 7.38 9.21 10.60
3 0.07 0.11 0.22 0.35 0.58 6.25 7.81 9.35 11.34 12.84
4 0.21 0.30 0.48 0.71 1.06 7.78 9.49 11.14 13.28 14.86
5 0.41 0.55 0.83 1.15 1.61 9.24 11.07 12.83 15.09 16.75

SEHH2311 Foundations of Data Science Page 6

 Example 1
(a) If 𝑋𝑋~𝜒𝜒 2 (10), find 𝑤𝑤 such that 𝑃𝑃 𝑋𝑋 ≤ 𝑤𝑤 = 0.975
(b) If 𝑋𝑋~𝜒𝜒 2 6 , find 𝑤𝑤 such that 𝑃𝑃 𝑋𝑋 > 𝑤𝑤 = 0.95
2
(c) Find 𝜒𝜒0.01;(12) .
2
(d) Find 𝜒𝜒0.975;(12) .
Solution:
(a) 𝑤𝑤 = 20.48
(b) 𝑤𝑤 = 1.64
2
(c) 𝜒𝜒0.01; 12 = 26.22
2
(d) 𝜒𝜒0.975; 12 = 4.40

SEHH2311 Foundations of Data Science Page 7

 Example 2
(a) If 𝑋𝑋~𝜒𝜒 2 (9), find 𝑤𝑤 such that 𝑃𝑃 𝑋𝑋 ≤ 𝑤𝑤 = 0.05
(b) If 𝑋𝑋~𝜒𝜒 2 13 , find 𝑤𝑤 such that 𝑃𝑃 𝑋𝑋 > 𝑤𝑤 = 0.9
2
(c) Find 𝜒𝜒0.025;(3) .
2
(d) Find 𝜒𝜒0.995;(10) .
Solution:

SEHH2311 Foundations of Data Science Page 8

 Example 2
(a) If 𝑋𝑋~𝜒𝜒 2 (9), find 𝑤𝑤 such that 𝑃𝑃 𝑋𝑋 ≤ 𝑤𝑤 = 0.05
(b) If 𝑋𝑋~𝜒𝜒 2 13 , find 𝑤𝑤 such that 𝑃𝑃 𝑋𝑋 > 𝑤𝑤 = 0.9
2
(c) Find 𝜒𝜒0.025;(3) .
2
(d) Find 𝜒𝜒0.995;(10) .
Solution:
(a) 𝑤𝑤 = 3.33
(b) 𝑤𝑤 = 7.04
2
(c) 𝜒𝜒0.025; 3 = 9.35
2
(d) 𝜒𝜒0.995; 10 = 2.16

SEHH2311 Foundations of Data Science Page 9

 F Distribution
• Suppose 𝑌𝑌1 and 𝑌𝑌2 are independent random variables, such that 𝑌𝑌1 ~𝜒𝜒 2 (𝑑𝑑1 ),
𝑌𝑌2 ~𝜒𝜒 2 (𝑑𝑑2 )
𝑌𝑌1 /𝑑𝑑1
• If 𝑋𝑋 = , then 𝑋𝑋 is said to have a 𝐹𝐹-distribution with 𝑑𝑑1 and 𝑑𝑑2 degrees
𝑌𝑌2 /𝑑𝑑2
of freedom. i.e. 𝑋𝑋~𝐹𝐹(𝑑𝑑1 , 𝑑𝑑2 )
• 𝑑𝑑1 is the numerator d.f. and 𝑑𝑑2 is the denominator d.f.
• pdf of 𝐹𝐹-distribution
𝑑𝑑1 𝑑𝑑1 +𝑑𝑑2
𝑑𝑑1 −
1 𝑑𝑑1 2
2 −1
𝑑𝑑1 2
𝑓𝑓 𝑥𝑥 = 𝑥𝑥 1 + 𝑥𝑥 , 0 ≤ 𝑥𝑥 < ∞
𝑑𝑑 𝑑𝑑 𝑑𝑑2 𝑑𝑑2
𝐵𝐵 21 , 22

where 𝐵𝐵(𝑢𝑢1 , 𝑢𝑢2 ) is the Beta function.

• pdf of F-distribution is skewed to the right.
• CDF of 𝐹𝐹-distribution
𝑥𝑥
𝐹𝐹 𝑥𝑥 = � 𝑓𝑓 𝑢𝑢 𝑑𝑑𝑑𝑑 (This is somewhat complicated)
0

SEHH2311 Foundations of Data Science Page 10

 F Distribution
𝑑𝑑2
• 𝐸𝐸 𝑋𝑋 = , for 𝑑𝑑2 > 2.
𝑑𝑑2 −2
– When 𝑑𝑑2 ≤ 2, 𝐸𝐸(𝑋𝑋) doesn’t exist

2𝑑𝑑22 (𝑑𝑑1 +𝑑𝑑2 −2)

• 𝑉𝑉𝑉𝑉𝑉𝑉 𝑋𝑋 = , for 𝑑𝑑2 > 4
𝑑𝑑1 𝑑𝑑2 −2 2 (𝑑𝑑2 −4)
– When 𝑑𝑑2 ≤ 4, 𝑉𝑉𝑉𝑉𝑉𝑉(𝑋𝑋) doesn’t exist

SEHH2311 Foundations of Data Science Page 11

 pdf of F-Distribution
pdf of F-Distributions with different degrees of freedom

f(x)

SEHH2311 Foundations of Data Science Page 12

 CDF of F-Distribution
CDF of F-Distributions with different degrees of freedom

F(x)

SEHH2311 Foundations of Data Science Page 13

 Using F-Distribution Table
F-Distribution tables are often used F-Distribution with 𝑑𝑑1 = 3 and 𝑑𝑑2 = 5
when conducting hypothesis testing.

Example:
If 𝑋𝑋~𝐹𝐹(3,5), then 𝑃𝑃 𝑋𝑋 > 12.06 = 0.01
or
𝐹𝐹0.01; 3,4 = 12.06

F-Distribution Table (𝛼𝛼 = 0.01)

d1
d2 1 2 3 4 5 6 7 8 9 10 ∞
1 4052.18 4999.50 5403.35 5624.58 5763.65 5858.99 5928.36 5981.07 6022.47 6055.85 6365.86
2 98.50 99.00 99.17 99.25 99.30 99.33 99.36 99.37 99.39 99.40 99.50
3 34.12 30.82 29.46 28.71 28.24 27.91 27.67 27.49 27.35 27.23 26.13
4 21.20 18.00 16.69 15.98 15.52 15.21 14.98 14.80 14.66 14.55 13.46
5 16.26 13.27 12.06 11.39 10.97 10.67 10.46 10.29 10.16 10.05 9.02

SEHH2311 Foundations of Data Science Page 14

 Example 3
(a) If 𝑋𝑋~𝐹𝐹 3,10 , find 𝑤𝑤 such that 𝑃𝑃 𝑋𝑋 ≤ 𝑤𝑤 = 0.99
(b) If 𝑋𝑋~𝐹𝐹 5,15 , find 𝑤𝑤 such that 𝑃𝑃 𝑋𝑋 > 𝑤𝑤 = 0.025
(c) Find 𝐹𝐹0.05;(7,60) .
(d) Find 𝐹𝐹0.01;(30,10) .

Solution:
(a) 𝑤𝑤 = 6.55
(b) 𝑤𝑤 = 3.58
(c) 𝐹𝐹0.05; 7,60 = 2.17
(d) 𝐹𝐹0.01; 30,10 = 4.25

SEHH2311 Foundations of Data Science Page 15

 Example 4
(a) If 𝑋𝑋~𝐹𝐹 6,12 , find 𝑤𝑤 such that 𝑃𝑃 𝑋𝑋 ≤ 𝑤𝑤 = 0.975
(b) If 𝑋𝑋~𝐹𝐹 2,20 , find 𝑤𝑤 such that 𝑃𝑃 𝑋𝑋 > 𝑤𝑤 = 0.01
(c) Find 𝐹𝐹0.025;(10,30) .
(d) Find 𝐹𝐹0.05;(8,120) .

Solution:
(a)
(b)
(c)
(d)

SEHH2311 Foundations of Data Science Page 16

 Example 4
(a) If 𝑋𝑋~𝐹𝐹 6,12 , find 𝑤𝑤 such that 𝑃𝑃 𝑋𝑋 ≤ 𝑤𝑤 = 0.975
(b) If 𝑋𝑋~𝐹𝐹 2,20 , find 𝑤𝑤 such that 𝑃𝑃 𝑋𝑋 > 𝑤𝑤 = 0.01
(c) Find 𝐹𝐹0.025;(10,30) .
(d) Find 𝐹𝐹0.05;(8,120) .

Solution:
(a) 𝑤𝑤 = 3.73
(b) 𝑤𝑤 = 5.85
(c) 𝐹𝐹0.025; 10,30 = 2.51
(d) 𝐹𝐹0.05; 8,120 = 2.02

SEHH2311 Foundations of Data Science Page 17

 t-Distribution
t-Distribution is also called Student's t-distribution, but why do we
need t-distribution?

𝑋𝑋−𝜇𝜇
If 𝑋𝑋 is 𝑁𝑁(𝜇𝜇, 𝜎𝜎 2 ), we know that 𝑍𝑍 = is 𝑁𝑁(0,1). i.e. standard
𝜎𝜎
score of normal random variables will distributed as standard
normal.

In many practical situations, 𝜎𝜎 is unknown. What will happen if we

𝑋𝑋−𝜇𝜇
replace it by sample standard deviation 𝑠𝑠? (i.e. )
𝑠𝑠

𝑋𝑋−𝜇𝜇
Will be standard normal or something else?
𝑠𝑠

SEHH2311 Foundations of Data Science Page 18

 t-Distribution
• If 𝑌𝑌 and 𝑍𝑍 are independent random variables such that 𝑌𝑌~𝜒𝜒 2 (𝜈𝜈) and
𝑍𝑍~𝑁𝑁(0,1).
𝑍𝑍
• If we let 𝑋𝑋 = , then 𝑋𝑋 is said to have a Student’s t Distribution with 𝜈𝜈
𝑌𝑌/𝜈𝜈
degrees of freedom. i.e 𝑋𝑋~𝑡𝑡(𝜈𝜈).
• pdf of t-distribution
𝜈𝜈 + 1 −
𝜈𝜈+1
Γ 𝑥𝑥 2 2
𝑓𝑓 𝑥𝑥 = 2
𝜈𝜈 1 + 𝜈𝜈 , −∞ < 𝑥𝑥 < ∞
𝜈𝜈𝜈𝜈Γ
2
where Γ(𝑢𝑢) is the gamma function.
• CDF of t-distribution
𝑥𝑥
𝐹𝐹 𝑋𝑋 = ∫−∞ 𝑓𝑓 𝑢𝑢 𝑑𝑑𝑑𝑑,−∞ < 𝑥𝑥 < ∞

SEHH2311 Foundations of Data Science Page 19

 t-Distribution
• If 𝑌𝑌 and 𝑍𝑍 are independent random variables such that 𝑌𝑌~𝜒𝜒 2 (𝜈𝜈) and
𝑍𝑍~𝑁𝑁(0,1).
𝑍𝑍
• If we let 𝑋𝑋 = , then 𝑋𝑋 is said to have a Student’s t Distribution with 𝜈𝜈
𝑌𝑌/𝜈𝜈
degrees of freedom. i.e 𝑋𝑋~𝑡𝑡(𝜈𝜈).
• pdf of t-distribution
𝜈𝜈 + 1 −
𝜈𝜈+1
Γ 𝑥𝑥 2 2
𝑓𝑓 𝑥𝑥 = 2
𝜈𝜈 1 + 𝜈𝜈 , −∞ < 𝑥𝑥 < ∞
𝜈𝜈𝜈𝜈Γ
2
where Γ(𝑢𝑢) is the gamma function.
• pdf of t-distribution is symmetric about 0, very similar to normal. However it
has heavier tails.
• CDF of t-distribution
𝑥𝑥
𝐹𝐹 𝑋𝑋 = ∫−∞ 𝑓𝑓 𝑢𝑢 𝑑𝑑𝑑𝑑,−∞ < 𝑥𝑥 < ∞

SEHH2311 Foundations of Data Science Page 20

 t-Distribution
• 𝐸𝐸 𝑋𝑋 = 0 for 𝜈𝜈 > 1
– when 𝜈𝜈 = 1, 𝐸𝐸(𝑋𝑋) doesn’t exist
𝜈𝜈
• 𝑉𝑉𝑉𝑉𝑉𝑉 𝑋𝑋 = for 𝜈𝜈 > 2
𝜈𝜈−2
– when 𝜈𝜈 ≤ 2, 𝑉𝑉𝑉𝑉𝑉𝑉(𝑋𝑋) doesn’t exist
• When 𝜈𝜈 goes to infinity, the distribution of 𝑋𝑋 approaches the
standard normal distribution 𝑁𝑁(0,1).

SEHH2311 Foundations of Data Science Page 21

 pdf of t-distribution
pdf of t-distribution with different degrees of freedom

SEHH2311 Foundations of Data Science Page 22

 CDF of t-distribution
CDF of t-distribution with different degrees of freedom

SEHH2311 Foundations of Data Science Page 23

 Using t-Distribution Table
t-Distribution table is often used when t-Distribution with 4 degrees of freedom
constructing confidence intervals and carrying
out hypotheses tests.
Example:
If 𝑋𝑋~𝑡𝑡(4), then

𝑃𝑃 𝑋𝑋 ≤ 2.776 = 0.975
or
𝑡𝑡0.025; 4 = 2.776

F(x) 0.600 0.650 0.700 0.750 0.800 0.850 0.900 0.950 0.975 0.990 0.995
α =1-F(x) 0.400 0.350 0.300 0.250 0.200 0.150 0.100 0.050 0.025 0.010 0.005
d.f.
1 0.325 0.510 0.727 1.000 1.376 1.963 3.078 6.314 12.706 31.821 63.657
2 0.289 0.445 0.617 0.816 1.061 1.386 1.886 2.920 4.303 6.965 9.925
3 0.277 0.424 0.584 0.765 0.978 1.250 1.638 2.353 3.182 4.541 5.841
4 0.271 0.414 0.569 0.741 0.941 1.190 1.533 2.132 2.776 3.747 4.604
5 0.267 0.408 0.559 0.727 0.920 1.156 1.476 2.015 2.571 3.365 4.032

SEHH2311 Foundations of Data Science Page 24

 Example 5
(a) If 𝑋𝑋~𝑡𝑡(9), find 𝑤𝑤 such that 𝑃𝑃 𝑋𝑋 ≤ 𝑤𝑤 = 0.75.
(b) If 𝑋𝑋~𝑡𝑡(9), find 𝑤𝑤 such that 𝑃𝑃 𝑋𝑋 ≤ 𝑤𝑤 = 0.25.
(c) If 𝑋𝑋~𝑡𝑡(25), find 𝑤𝑤 such that 𝑃𝑃 𝑋𝑋 > 𝑤𝑤 = 0.1.
(d) Find 𝑡𝑡0.005;(15).
(e) Find 𝑡𝑡0.99; 6 .
Solution:
(a) 𝑤𝑤 = 0.703
(b) 𝑤𝑤 = −0.703
(c) 𝑤𝑤 = 1.316
(d) 𝑡𝑡0.005; 15 = 2.947
(e) 𝑡𝑡0.99; 6 = −𝑡𝑡0.01 6 = −3.143

SEHH2311 Foundations of Data Science Page 25

 Example 6
(a) If 𝑋𝑋~𝑡𝑡(10), find 𝑤𝑤 such that 𝑃𝑃 𝑋𝑋 ≤ 𝑤𝑤 = 0.9.
(b) If 𝑋𝑋~𝑡𝑡(12), find 𝑤𝑤 such that 𝑃𝑃 𝑋𝑋 > 𝑤𝑤 = 0.99.
(c) If 𝑋𝑋~𝑡𝑡(8), find 𝑤𝑤 such that 𝑃𝑃 𝑋𝑋 > 𝑤𝑤 = 0.2.
(d) Find 𝑡𝑡0.8;(11) .
(e) Find 𝑡𝑡0.1; 13 .

Solution:
(a)
(b)
(c)
(d)
(e)

SEHH2311 Foundations of Data Science Page 26

 Example 6
(a) If 𝑋𝑋~𝑡𝑡(10), find 𝑤𝑤 such that 𝑃𝑃 𝑋𝑋 ≤ 𝑤𝑤 = 0.9.
(b) If 𝑋𝑋~𝑡𝑡(12), find 𝑤𝑤 such that 𝑃𝑃 𝑋𝑋 > 𝑤𝑤 = 0.99.
(c) If 𝑋𝑋~𝑡𝑡(8), find 𝑤𝑤 such that 𝑃𝑃 𝑋𝑋 > 𝑤𝑤 = 0.2.
(d) Find 𝑡𝑡0.8;(11) .
(e) Find 𝑡𝑡0.1; 13 .

Solution:
(a) 𝑤𝑤 = 1.372
(b) 𝑤𝑤 = −2.681
(c) 𝑤𝑤 = 0.889
(d) 𝑡𝑡0.8; 11 = −0.876
(e) 𝑡𝑡0.1; 13 = 1.350

SEHH2311 Foundations of Data Science Page 27

 Motivations of 𝒕𝒕, 𝝌𝝌𝟐𝟐 and 𝑭𝑭 Distributions
• The distributions in this lectures are motivated by random
variables encountered when estimating unknown population
parameters.
∑𝑛𝑛 � 2
𝑖𝑖=1 𝑋𝑋𝑖𝑖 −𝑋𝑋
• Sample variance 𝑠𝑠 2 = is related to 𝜒𝜒 2 distribution
𝑛𝑛−1
�
𝑋𝑋−𝜇𝜇
• t-distribution is related to 𝑡𝑡 = 𝑠𝑠 , which is a standard score
𝑛𝑛
with population variance replaced by sample variance.
• F-distribution is used in the comparison of sample variances of
two populations

SEHH2311 Foundations of Data Science Page 28