Probability and Statistics

Lecture 1 - Introduction

Adina Amanbekkyzy
[email protected]
Lecture overview:

1. The primary goal of mathematical statistics and

modelling
2. Basic definitions: types of statistics, types of
variables, types of datasets, type of
measurement scales
3. Measures of central tendency: mean, mode,
median
4. Measures of dispersian: IQR, variance, standard
deviation, skewness

2
1. Introduction
• Statistics – the art of learning from data. It’s
concerned with the collection of data, its
subsequent description, and its analysis, which
leads to the drawing of conclusions.

• Descriptive statistics – part of statistics, concerned

with the description and summarization of data.

• Inferential statistics – concerned with the drawing

of conclusions. To be able to draw logical
conclusions from data probability models should be
created
3
Basic cycles of probabilistic modeling
and analysis
• Mathematical models are descriptions of situations
in the real world and are usually written to help
solve some kind of problem.

• Examples: business (production of USB, computer

chips), medicine (vaccines creation, mass
production), government (laws, regulations).

4
 Flowchart. The modelling process
1. Recognising a real-world problem

2. Making the mathematical model

3. Use the model to make some predictions

4. Collect some real-world data

5. Compare the predictions with the real-world data

6. Test whether the model ’works’

7. Improve the model

5
Probabilistic modeling and analysis

6
Key terms
• Population – a total collection of elements. (For
instance, all residents of a given state, all the
television set produced in the last year by a
particula manufacturer)

• Sample – a subgroup of population elements to

learn about it.
Key terms
• From the sample data, we can calculate a
statistic. A statistic is a number that
represents a property of the sample.

• A parameter is a number that is a property of

the population.

• A variable (𝑋, 𝑌, 𝑒𝑡𝑐. ) is a characteristic of

interest for each person orthin in a population
 Types of variables
A lot of the subject of statistics involves analysing data.

Data consists of a number of observations (or

measurements).

Each of these observations records the value of a

particular variable.

There are different kinds of variables.

9
 Types of variables
• Qualitative variables (categorical) - variables
that take non-numerical values (i.e. they're
not numbers)

• Quantitative variables (numerical) -

variables that take numerical values (i.e.
they're numbers)

10
 Example
An employer collects information about the computers in
his office. He gathers observations of the 5 variables
shown in the table.
1. Manufacturer Bell Banana Deucer Deucer

2. Processor speed
2.6 2.1 1.8 2.2
(in GHz)
3. Year of purchase 2009 2010 2011 2009
4. Memory (in MB) 2 3 3.1 4.8
5. Colour Grey Grey Grey Black

Which of the 5 variables are a) qualitative?

b) quantitative?
11
Answers
a) The variables 'Manufacturer' and 'Colour' take
values that are not numbers. So there are two
qualitative variables: 'Manufacturer' and 'Colour‘.

b) The variables 'Processor speed', 'Year of purchase'

and 'Memory‘ take values that are numbers. So there
are three quantitative variables: 'Processor speed' ,
'Year of purchase' and 'Memory'.

12
Self-study
A mechanic collects the following information about
cars he services:
Make, Mileage, Colour, Number of doors, Cost of
service

Write down all the variables from this list that are:
a) qualitative
b) quantitative

13
Types of variables
• Qualitative variables

• Quantitative variables
• Discrete
• Continuous

14
Types of variables
There are then two different types of quantitative
variables.
• A discrete variable can only take certain values
within a particular range (e.g. shoe sizes) - this
means there are 'gaps' between possible values (you
can't take size 9.664 shoes, for example)

• A continuous variable can take any value within a

particular range (e.g. lengths, time or masses) - there
are no gaps between possible values.

15
Example
The variables below are all quantitative.
(i) length, (ii) weight, (iii) number of brothers, (iv)
time, (v) total value of 6 coins from down in the back
of my sofa

a) Which of the above quantitative variables are

continuous?
b) Which of the above quantitative variables are
discrete?

16
Answers
a) 'Length', 'weight' and 'time' can all take any value
in a range. So the continuous variables are:
‘length', 'weight' and 'time'.

b) 'Number of brothers' and 'total value of 6 coins'

can only take certain values. So there are two
discrete variables - these are: ‘number of
brothers and 'total value of 6 coins'.

17
Types of data sets
Data is often shown in the form of a table.
1. Frequency tables show the number of
observations of various values.

For example, this frequency table shows the number of

bananas in thirty 1.5 kg bags.
Number of bananas 8 9 10 11 12
Frequency 3 7 10 6 4
18
Types of data sets
2. Grouped frequency tables show the number of
observations whose values fall within certain classes
(i.e. ranges or groups of values).
They're often used when there is a large range of
possible values.

Number of potatoes 50-55 56-60 61-65 66-70 71-75

Frequency 1 8 12 7 2
Note: grouped frequency tables don't tell you the
exact value of the observations - just the most and
the least they could be.

19
Types of measurement scales

This is a simple case when k=1

!"#$%#&'(
frequency density =
')*++ ,-./0
20
 Histograms

Histogram

A histogram is a graphical representation of the

distribution of numerical data.

21
 Bar charts
Histogram vs. Bar chart

Difference between Histogram and Bar chart

1. Shows distributions of variables 1. Used to compare variables
2. Plots binned quantitative data 2. Plots categorical data
3. Show distribution of data 3. Bars can be reordered in bar
(symmetrical, positively, negatively charts but not in histograms
skewed), Bar chart can’t show 22
Skewness
Negative skewed Symmetrical Positive skew

23
Example
Here’s some data showing the heights of 24 people:

24
 Here’s some data showing the heights of 24 people:

Correct histogram
Incorrect
histogram

25
 Draw and interpret histograms
(i) The vertical axis shows frequency density.
(ii) The horizontal axis has a continuous scale like an
ordinary graph and there are no gaps between the
columns.
(iii) A bar's left-hand edge corresponds to the lower class
boundary. A bar's right-hand edge corresponds to the
upper class boundary.

On a histogram, the frequency in a class is proportional

to the area of its bar. In other words,
frequency = k x area of bar (where k is a number).
26
Example
This histogram shows the heights of a group of
people. There are 6 people between 155 and 160 cm
tall.

How many people in the group are between 130

cm and 155 cm tall?
27
Solution
1. Before looking at the 130-155 class, you need to
use the information you have about the 155-160
class.

Work out the area of the Width of bar=160-155 = 5

bar for 155 cm to 160 cm Height of bar = 0.6
to find how area and So area of bar = 5 x 0.6 = 3
frequency are related.

2. An area of 3 represents a frequency of 6 - so

'frequency = 2 x area'
28
 Class Frequency Area
155-160cm 6 3

Since, in this particular example k1 =2

• frequency = k1 x area of bar=
=2 x area of bar; 6=2*3
or 𝐒𝐜𝐚𝐥𝐞 1:2
• frequency x k2= frequency *1/2 = 𝑨𝒓𝒆𝒂𝒉𝒊𝒔𝒕𝒐𝒈𝒓𝒂𝒎 : 𝒇𝒓𝒆𝒒𝒖𝒆𝒏𝒄𝒚𝒑𝒆𝒐𝒑𝒍𝒆

=area of bar; 6*1/2=3

𝑨𝒓𝒆𝒂𝒂𝒄𝒕𝒖𝒂𝒍 = 𝑨𝒓𝒆𝒂𝒉𝒊𝒔𝒕𝒐𝒈𝒓𝒂𝒎 ∗𝐤=
IMPORTANT: =𝟏 ∗ 𝟐 = 𝒇𝒓𝒆𝒒𝒖𝒆𝒏𝒄𝒚𝒑𝒆𝒐𝒑𝒍𝒆

1)Simply to say we can scale in both 𝐒𝒐, 𝒊𝒏 𝒐𝒕𝒉𝒆𝒓 𝒘𝒐𝒓𝒅𝒔, 𝒘𝒆 𝒏𝒆𝒆𝒅 𝒕𝒐

directions 𝒎𝒖𝒍𝒕𝒊𝒑𝒍𝒚 𝒕𝒉𝒊𝒔 𝒉𝒊𝒔𝒕𝒐𝒈𝒓𝒂𝒎 by 2,
or enlarge it twice. So that
2) k1 and k2 are reciprocals→ k1=1/ k2 Area=frequency
29
 Class Frequency Area
155-160cm 6 3

Since in this particular example k1 =2

• frequency = k1 x area of bar = 2 x area of bar; 6=2*3

or
• frequency x k2= frequency *1/2 = area of bar; 6*1/2=3

IMPORTANT:
1)Simply to say we can scale in both directions
2) k1 and k2 are reciprocals→ k1=1/ k2
30
Solution continued
3. Now you need to find the area of the bar from 130
cm to 155 cm:

Width of bar= 155 -130=25

Height of bar = 0.2
So area of bar = 25 x 0.2 = 5

Since ‘frequency = 2 x area’. So 10 people are

between 130 cm and 155 cm tall.

31
Self-study
For the same histogram

how many people in the group are over 165 cm tall?

32
Solution

33
 Stem and leaf diagram
1) What are Stem and leaf diagrams?

Stem and leaf diagrams are another way to represent data. In

fact, the data values themselves appear in a stem and leaf
diagram.

2) Why do we need Stem and leaf diagrams?

Although a histogram shows how observations are distributed

across groups, it does not show the exact values of individual
observations. A different kind of graphical display, called a Stem
and leaf plot, shows the exact values of individual observations.
 Stem and leaf diagram
• Each data value is split into a 'stem' and a ‘leaf'.
• A complete stem and leaf diagram looks something
like this:

A stem and leaf diagram always needs a key to

tell you how to read it.
Measures of central tendency
∑ EF
• Mean: 𝑥̅ = ∑E
=𝜇
• Mode: modal value - most frequently occurring
data value
Note:
1) No mode if each data value occurs only once;
2) Every value that occurs the most frequently is a
mode.
• Median: value in the middle of the dataset when all
the data values are placed in order of size
Example
Example cont.
c) Find the mean number of the letters.
∑ 𝑓𝑥 213
𝑥̅ = = 𝜇= = 2.13
∑𝑓 100
Example
 Frequency tables for
grouped data
Mass of potato (m, in g) Frequency

16

14

5
 Clacss width and midpoint

class width = upper class boundary –

lower class boundary

mid-point =
!"#$% &!'(( )"*+,'%- .*//$% &!'(( )"*+,'%-
0
 Example
This grouped frequency table shows the lengths (to
the nearest cm) of the same 50 potatoes.
Length of potato (l, in cm) Frequency The shortest potato that could go in
the 6-7 class would actually have a
4–5 5 length of 5.5 cm (since 5.5 cm is
6–7 11 rounded up to 6 cm when measuring
8–9 15 to the nearest cm). So the lower class
10 – 11 16 boundary of the 6-7 class is 5.5 cm.

12 – 13 3

The upper class boundary of the 6-7 class is the same as the
lower class boundary of the 8-9 class - this is 7.5 cm. This
means there are never any gaps between classes.
42
 Example
This table shows the data for all groups:
Length of potato Lower class Upper class Class
Frequency Midpoint
(l, in cm) boundary boundary width

4–5 5 3.5 5.5 2 4.5

6–7 11 5.5 7.5 2 6.5
8–9 15 7.5 9.5 2 8.5
10 – 11 16 9.5 11.5 2 10.5

12 – 13 3 11.5 13.5 2 12.5

43
 Example
The heights of a number of trees were recorded. The
data collected is shown in this table.

Height of trees to 0 - 5.5 5.5 - 10.5 10.5 – 15.5 15.5 – 20.5

nearest dm [0; 5.5) [5.5; 10.5) [10.5; 15.5) [15.5; 20.5)

Number of trees 26 17 11 6

Find an estimate of the mean height of the trees.

44
Solution
1) It’s best to make another table. Include extra
rows showing:

(i) The class mid-points (𝑥),

(ii) The values of 𝑓𝑥, where 𝑓 is the frequency,
(iii) Add an extra column for the totals ∑ 𝑓 and ∑ 𝑓𝑥.

45
 Solution
Height of
0 - 5.5 5.5 - 10.5 10.5 – 15.5 15.5 – 20.5
trees to
[0; 5.5) [5.5; 10.5) [10.5; 15.5) [15.5; 20.5)
Total
nearest dm

2.75 8 13 18
Number of
26 17 11 6
trees, f

71.5 136 143 108

2) Use the mean formula:

∑ EF GHI.H
𝑥̅ = ∑E
= KL
= 7.64 (to 2 d.p.)

46
Median. Example
To find an estimate for the median, use linear
interpolation. The table below shows the ‘tree data’
from the last example with the cumulative
frequency. Estimate the median height (to 2 d.p.).

Height of trees to 0 - 5.5 5.5 - 10.5 10.5 – 15.5 15.5 – 20.5

nearest m [0; 5.5) [5.5; 10.5) [10.5; 15.5) [15.5; 20.5)

Number of trees 26 17 11 6

Cumulative
26 43 54 60
frequency
47
 Median. Solution
1. First, find which class the median is in. Since ( = *+
= 30, there
) )
are 30 values less than or equal to the median. This means the
median must be in the 5.5-10.5 class.

2.

48
 Solution
3.

4.

5.

49
Median
6.

50
 Median. Shorter equation

OR
𝑎M
𝑚 = 𝑙𝑜𝑤𝑒𝑟 𝑐𝑙𝑎𝑠𝑠 𝑏𝑜𝑢𝑛𝑑𝑎𝑟𝑦 + 𝑐𝑙𝑎𝑠𝑠 𝑤𝑖𝑑𝑡ℎ×
𝑏M
where 𝑎M , 𝑏M are the proper differences between
cumulative frequencies.
51
Example
Estimate the median length of the
newts described in this table (round
to 1 d.p.)

52
Solution

53
Exercise

54
 Solution
a) 0-2 letters. All the classes are the same width, so you
can use the frequency to find the modal class (instead of
the frequency density).
b) First we add some extra columns to the table:
Number of
letters

0-2 20 1 20 $ 𝑓 = 50, $ 𝑓𝑥 = 209

3-5 16 4 64 So estimate of mean = 209÷
6-8 7 7 49 50 = 4.18
9-11 5 10 50
12-14 2 13 26
55
Solution continued
c) Since ∑ 𝑓 ÷ 2 = 50 ÷ 2 = 25, the median is
halfway between the values in this position (25) and
the next position (26) in ordered list. So the median
must be in the class 3-5.

56
Comparing measures of central
tendency

Which one is the most suitable to

describe data?
Mean?
Median?
Mode?

57
1) What type of data we are dealing with?

Mean - quantitative data

Median - quantitative data
Mode – qualitative data

58
1) Do we have outliers (a very large or small
numbers that differs dramatically from
main data set)?

If yes then
Mean - quantitative data (not suitable)
Median - quantitative data (suitable)
59
 Mean
Advantage:
• The mean is a good average because you use
all your data in working it out.
Disadvantage:
• It can be heavily affected by extreme
values/outliers and by distributions of data
values that are not symmetric.
• And it can only be used with quantitative data
(i.e. numbers).
60
 Median
Advantage:
• The median is not affected by extreme values
- so this is a good average to use when you have
outliers.
• This also makes it a good average to use when the
data set is not symmetric.

• Disadvantage:
• It can be used with quantitative data (i.e. numbers).
61
 Mode
Advantage:
• The mode can be used with qualitative (non-
numerical) data.

Disadvantage:
• Some data sets can have more than one
mode (and if every value in a data set occurs
just once, then the mode isn't helpful at all).
62
Quick question!

63
Quick question!

64
Quick question!

65
 Solution

Median? Mean?

a) Median – most employees will earn relatively Iow

salaries but a few may earn much higher salaries, so
the mean could be heavily affected by these few
high salaries (outliers).

66
Solution

Median? Mean?

b) Mean - the data should be reasonably symmetrical so

the mean would be a good measure of location. The
median would be good as well (for a symmetric data
set, it should be roughly equal to the mean).

67
Solution

c) Mode - make of car is qualitative data so the mode is

the only average that can be found.

68
 Solution

d) Mean - the data should be reasonably symmetrical so

the mean would be a good measure of location. The
median would be good as well (for a symmetric data
set, it should be roughly equal to the mean).

69
Solution

e) Median - most employees will perhaps travel fairly

short distances to work but a few employees may live
much further away (outliers). The median would not be
affected by these few high values.

70
Measures of dispersian
• Quartiles Q1, Q2, Q3 split the data into four parts
where Q1 is the lower quartile, Q2 is the median, Q3 is the
upper quartile.

Interquartile range (IQR) = Q3 - Q1

𝒙)𝟐
∑(𝒙PQ ∑ 𝒙𝟐 ∑𝒙 𝟐 ∑ 𝒙𝟐
𝑽𝒂𝒓𝒊𝒂𝒏𝒄𝒆 = = − = − 𝒙𝟐 =
𝒏 𝒏 𝒏 𝒏
𝑺𝒕𝒂𝒏𝒅𝒂𝒓𝒅 𝒅𝒆𝒗𝒊𝒂𝒕𝒊𝒐𝒏 = 𝑽𝒂𝒓𝒊𝒂𝒏𝒄𝒆 = 𝝈
 Example
Find the range of 7, 9, 4, 6, 3, 2, 8, 1, 10, 15, 11

1st step: Put the data in ascending order:

1, 2, 3, 4, 6, 7, 8, 9, 10, 11, 15

2nd step: Compute the range:

Range = 15 – 1 =14

72
 Exercise
Given the data set; -5, 17, 33, 22, 0.27, p
is known to have a range of 43.25, what
are the two possible missing values, p?
Possible missing values of p
-10.25 -5 0.27 17 22 33 38.25
Range = p = 38.25- (-5)=43.25
Range = p = 33-(-10,25)=43.25
However, we have better INTERQUARTILE
tool to measure dispersion RANGE !
73
 Interquartile range (IQR)
A more useful way to measure dispersion is to use the
interquartile range – but first we need to find the
quartiles.
Quartiles
Quartiles Q1, Q2, Q3 split the data into four parts

where Q1 is the lower quartile, Q2 is the median, Q3 is the upper

quartile.
Interquartile range

74
 Calculation of the Quartiles
(Discrete data)
• To calculate the lower quartile, Q1, divide n by 4. When n is a whole
4
number, find the midpoint between the corresponding term and the
n
term above. When 4 is not a whole number, round the number up and
pick the corresponding term.

• To calculate the upper quartile, Q3, divide n by 4 and multiply by 3.

3n
When is a whole number, find the midpoint between the
4
corresponding term and the term above. When 3n is not a whole
4
number, round the number up and pick the corresponding term.

Note: Don’t forget that n represents the number of elements in the

sample!
75
Example

IQR=2
76
Exercise

IQR=10-3=7

77
 Calculation of the Quartiles
(Continuous and grouped data)

78
 Example
The length of time, spent on the internet each evening by
a group of students is shown in the table below.
Find Q1 and Q3 and then the IQR for the data below.
For grouped data we need to use interpolation method.

Lower Quartile Q1
1st Step:
n 70
Q1 : = = 17.5th
4 4
2nd step:
79
 Example
3rd step:

80
 Example

Upper Quartile Q3
1st Step:
3n 3 * 70
Q3 : = = 52.5th
4 4

2nd step:
3rd step:

Interquartile range (IQR)

81
 Exercise
The weights of 31 Jersey cows were recorded to the
nearest kilogram. The weights are shown in the table.
a. Complete the cumulative frequency column in the table.
Weight of a cattle Frequency
(kg)
b. Find Q1. 300-349 3
350-399 6
c. Find Q3. 400-449 10
450-499 7
500-549 5

d. Work out the interquartile range.

82
 Exercise
The weights of 31 Jersey cows were recorded to the
nearest kilogram. The weights are shown in the table.
a. Complete the cumulative frequency column in the table.

Weight of a Frequency Cumulative

cattle (kg) frequency Data is
300-349 3 3
grouped,
350-399 6 9
thus use
400-449 10 19
450-499 7 26
interpolation
500-549 5 31
83
 Exercise
b. Find Q1.

𝟑𝟏
𝑸𝟏 = = 𝟕. 𝟕𝟓𝒕𝒉 𝒗𝒂𝒍𝒖𝒆, 𝒔𝒐 𝑸𝟏 𝒊𝒔 𝒊𝒏 𝒕𝒉𝒆 𝒄𝒍𝒂𝒔𝒔 𝟑𝟓𝟎 − 𝟑𝟗𝟗
𝟒

𝑸𝟏 − 𝟑𝟒𝟗. 𝟓 𝟕. 𝟕𝟓 − 𝟑
=
𝟑𝟗𝟗. 𝟓 − 𝟑𝟒𝟗. 𝟓 𝟗−𝟑

𝑸𝟏 = 39.58+349.5 ≈ 389.1

84
 Exercise
c. Find Q3.

31
𝑄-= 3 ∗ = 23.25th value
4
𝑄- 𝑖𝑠 𝑖𝑛 𝑡ℎ𝑒 𝑐𝑙𝑎𝑠𝑠 450 − 499
Q3 - 449.5 23.25 - 19
=
499.5 - 449.5 26 - 19
Q3 = 30.36 + 449.5 » 479.9
d. Work out the interquartile range.
IQR = Q3 - Q1 = 479.9 - 389.1 = 90.8
85
 Calculation of the Interpercentile range
Percentiles
Quartiles divide the data into four parts, where each part
contains the same number of data values. Percentiles are
similar, but they divide the data into 100 parts.

Interpercentile range

86
 Exercise
The height, in cm, of 70 seventeen year old boys were
recorded. Calculate:
a) The 90th percentile. b) The 10th percentile.
c) The 10% to 90% interpercentile range.

a) The 90th percentile

1st step:
2nd step: Interpolation method

3rd step:

87
 Exercise
b. The 10th percentile (try yourself!)
1st step:

2nd step: Interpolation method

3rd step:

c) The 10% to 90% interpercentile range.

88
 Determine if a value is an outlier
An OUTLIER is an extreme value that is a long way
from the majority of the readings in the data set.

89
 Example

3rd step: 30 is inside the lower fence, thus is not an outlier

210 is outside the upper fence, thus an outlier
90
 Variance and standard deviation
𝒙)𝟐
∑(𝒙PQ ∑ 𝒙𝟐 ∑𝒙 𝟐 ∑ 𝒙𝟐
𝑽𝒂𝒓𝒊𝒂𝒏𝒄𝒆 = = − = − 𝒙𝟐
𝒏 𝒏 𝒏 𝒏

𝑺𝒕𝒂𝒏𝒅𝒂𝒓𝒅 𝒅𝒆𝒗𝒊𝒂𝒕𝒊𝒐𝒏 = 𝑽𝒂𝒓𝒊𝒂𝒏𝒄𝒆

•Variance (𝜎 M ) and standard deviation (𝜎) are

measures of dispersion – they give an idea of how
spread out the data values are from the mean.

•The bigger the variance (or standard deviation),

the more spread out the data are.
 Example
The marks gained in a test by seven randomly selected
students are: 3, 4, 6, 2, 8, 8, 5
Calculate variance and standard deviation of the marks
of the seven students.

92
 Exercise
Find variance and standard deviation of the following
data set:
2, 3, 4, 4, 6, 11, 12

x 2 3 4 4 6 11 12
𝑥 ! 4 9 16 16 36 121 144
x = 2 + 3 + 4 + 4 + 6 + 11 + 12 = 42
"#$
Variance, 𝜎 ! = − 6! ≈ 13.4
%
Standard deviation, 𝜎 = 13.43 ≈3.7

93
References:
1. Palin A., Park A., Whiteley C., (2012), A-level
mathematics for Edexcel Statistics 1, CGP, UK.
2. Attwood, G., Clegg, A., Dyer, G. and Dyer, J
(2008), Edexcel AS and A-Level Modular
Mathematics series S2, Pearson, Harlow, UK.
3. Lecture notes, Statistics and Math for Life
Sciences courses, NUFYP, Nazarbayev
University.

94