Statistics 1 – Tutorial … 08/09/2020

Statistics 1 (MAT-15303) Tutorial 6: Normal distribution

Tutorial 6
Topics
Lecturer : Anja Boevé

 Empirical rule (68-95-99.7 rule) (video after T5)

 Normal distribution
 Standard normal distribution
 Transformations to Z-score and back
 Calculations for normal distributions with a table or calculator

Discrete random variable Continuous random variable

 Probability distribution for a discrete random variable y:

● Outcomes yi
A continuous random variable takes all values in an
interval of real values.
● Probabilities P(y = yi) = pi ● The probability distribution cannot be described by
● 0 ≤ pi ≤ 1 for every i individual outcomes yi with associated probabilities
● Σpi = 1 pi (as on the previous slide).

Example:
y = number of blue-eyed
persons in a group of
10 persons

Statistics 1 - Tutorial 6 3 Statistics 1 - Tutorial 6 4

Greenhouse Gases – car emissions Greenhouse Gases – car emissions

 Measurements CO2 by Environmental Protection

Agency (EPA)
 CO2 (grams / mile)
 5 types of cars

Statistics 1 - Tutorial 6 5 Statistics 1 - Tutorial 6 6

Handouts 1
Statistics 1 – Tutorial … 08/09/2020

Histogram CO2 BMW mini Continuous random variable

 A probability is a relative frequency in the long run,

 therefore the probability of A, P(A), is given as the
 Large number of area under the curve in interval A.
observations →
 bin size decreases →
 histogram becomes a
curve
Note: Total area under
● in other words a P(A)
probability the curve is 1.
density function

Statistics 1 - Tutorial 6 7 Statistics 1 - Tutorial 6 8

Pay attention!! Galton board

 clip galton board

https://www.youtube.com/watch?v=6YDHBFVIvIs
 For a discrete random variable:
● y = number of blue-eyed persons
● P(y ≤ 5 ) ≠ P(y < 5 )

 For a continuous random variable:

● y = birth weight
● P(y ≤ 3150) = P(y < 3150) !
● P(y = 3150) = 0 !

3150

Statistics 1 - Tutorial 6 9 Statistics 1 - Tutorial 6 10

Normal distribution Standard normal distribution

1 ( )
𝑓 𝑦 = ×𝑒
𝜎 2𝜋
 Z ~ N(0,1) σ=1

 Table 1 (inside front cover O&L)

 Symmetrical
 Unimodal 𝜎
μ=0

 Bell-shaped
 Notation: y ~ N(𝜇,𝜎)
is distributed as
Where
𝜇 = population mean
(expected value)
𝜎 = population standard deviation
𝜇 →𝑦

Statistics 1 - Tutorial 6 11 Statistics 1 - Tutorial 6 12

Handouts 2
Statistics 1 – Tutorial … 08/09/2020

Use of table standard normal distribution Transformation to a standard normal distribution

 Transformation: y ~ N(μ,σ) → Z ~ N(0,1)

y = μ + z σ , it follows z = (y-μ)/σ ‘z-score of y’

 Example: y ~ N(175,7.5)

Probability density p
P(y ≤ 165) =
P(175+ Z x 7.5 ≤ 165) = σ σ
P(Z ≤ (165-175)/7.5) =
P(Z ≤ -1.33) = Table 1 O&L
Examples: P(Z ≤ -3.05) = 0.0011 0.0918
→ z = -2.88 0
P(Z ≤ z) = 0.0020 µ y
y=µ + z σ
0 -1 0 1 Z
z=(y-μ)/σ
Statistics 1 - Tutorial 6 13 Statistics 1 - Tutorial 6 14

Exercises at the tutorial Exercise 4.53 (O&L, p.222)

Use Table 1 to find the area under the normal curve between
Make the following two exercises using both the table of these values: z is N(0,1) distributed, the table can be used directly!)
the standard normal distribution (table 1, O&L) and using a. z=0 and z=1.3
your graphing calculator (see Annexe of LN). Please,
always make a sketch of the normal distribution and b. z=0 and z=2.7
highlight the appropriate area under the curve. Let y be a normal random variable with μ=250 and σ=50. Find
the following:
 Exercise 4.53 (O&L, p.222 [6th edition: page 212. Note:
Replace z = 1.6 by z = 1.3 in item a) and z = 2.3 by z = a. P( y > 250)
2.7 in item b).] b. P(y > 150)
z~N(0,1) distributed, table 1 can be used directly! c. P(150 < y < 350)
 Exercise 4.65 (O&L, p.223 [6th edition: see LN]). d. k such that P(250-k < y < 250+k) = 0.60

Statistics 1 - Tutorial 6 15 Statistics 1 - Tutorial 6 16

Model for CO2 BMW - mini Calculation of probabilities and percentiles

 Calculate the probability that a randomly
 In Statistics 2 we will learn selected BMW-mini has a CO2 emission larger
than 320 grams/ mile.
how to check whether a P(y > 320) = ......
normal distribution is suitable. P(Z > (320-294)/26) =
 Here we assume: the normal P(Z > 1.00) = 0.1587
distribution is a good model  Calculate the probability that a randomly
for CO2 emission. selected BMW-mini has a CO2 emission
 N(μ, ) → μ and  unknown. between 270 and 320 grams/ mile
P(270 < y < 320) = ......
● Sample mean is an P((270-294)/26< Z <(320-294)/26) =
estimator for μ. n = 35
P(-0.92 <Z< 1.00)=P(Z<1.00)-P(Z<-0.92)=
● Sample standard deviation Mean = 294 grams/ mile 0.8413 - 0.1788 = 0.6625
is an estimator for . Standard deviation = 26 grams /mile Model CO2 emissions:
● The formulas for mean and  Calculate the 10th percentile of this
Model CO2 emissions: distribution: y ~ N(294, 26)
standard deviations are P(y < k) = 0.10
called estimators. y ~ N(294, 26) P(Z < (k-294)/26) = 0.10
(k-294)/26 = -1.28 so k = 260.7

Statistics 1 - Tutorial 6 17 Statistics 1 - Tutorial 6 18

Handouts 3
Statistics 1 – Tutorial … 08/09/2020

Calculation of desired expected value Graph and solution of the problem

 Due to new standards to reduce greenhouse gases the

limit for CO2 emissions for BMW-mini cars becomes 300
grams/ mile.
 Suppose that σ still equals 26 grams/mile and BMW P(y > 300) = 0.05
would like to produce Mini’s for which at most 5% of the
cars exceed this new limit of 300 grams/ mile. σ = 26, μ = ???
Furthermore assume that CO2 emission has a normal P(y > 300) = P(μ+zσ > 300) =
P(Z > (300-μ)/σ) = 0.05
distribution. What would be the desired (maximal) P(Z < (300-μ)/26) = 0.95
expected value for the CO2 emissions of BMW mini cars?
(300-μ)/26 = 1.645
 In short notation: y~ N(µ, 26), P(y > 300) = 0.05 µ
μ = 300 – 1.645*26 = 257.3

y~ N(µ, 26), P(y > 300) = 0.05

Statistics 1 - Tutorial 6 19 Statistics 1 - Tutorial 6 20

Learning objectives Tutorial 6 After the tutorial

After this tutorial you should be able to:

 Practice: Self-test after tutorial 6

 Know and apply the empirical rule (68-95-99.7 Exercises 6.1 and 6.2 in LN
rule) (video after T5)
Watch clip after tutorial 6.
 Recognize the normal distribution
 Recognize the standard normal distribution as a special  Reading: §4.12 O&L (pp. 190-194 – Central Limit
case Theorem will be explained in Tutorial 8) [6th
 Transform to z-score and back edition: pp. 181-185]
§14 and 15 of Supplementary Subject Matter
 Calculate fractions/probabilities/percentilesfor given in LN
value of Z/y, and vice versa, given a normally
distributed variable, with a table or calculator

21 Statistics 1 - Tutorial 6 22

Handouts 4