Chapter

1
Descriptive statistics

Contents
1.1 Statistical vocabulary . . . . . . . . . . . . . . . . . . . . . 2

1.2 Data description . . . . . . . . . . . . . . . . . . . . . . . . . 5

1.2.1 Tables . . . . . . . . . . . . . . . . . . . . . . . . . . . 5

1.2.2 Graphics . . . . . . . . . . . . . . . . . . . . . . . . . 6

1.3 Position parameters . . . . . . . . . . . . . . . . . . . . . . 8

1.3.1 Mean . . . . . . . . . . . . . . . . . . . . . . . . . . . 8

1.3.2 Mode . . . . . . . . . . . . . . . . . . . . . . . . . . . 9

1.3.3 Median . . . . . . . . . . . . . . . . . . . . . . . . . . 10

1.3.4 Quartiles . . . . . . . . . . . . . . . . . . . . . . . . . 12

1.4 Dispersion parameters . . . . . . . . . . . . . . . . . . . . . 13

1.4.1 Range . . . . . . . . . . . . . . . . . . . . . . . . . . . 13

1.4.2 Variance . . . . . . . . . . . . . . . . . . . . . . . . . 14

1.4.3 Standard deviation . . . . . . . . . . . . . . . . . . . 14

1.4.4 Coefficient of variation . . . . . . . . . . . . . . . . . 14

1.5 Shape parameter . . . . . . . . . . . . . . . . . . . . . . . . 15

1.5.1 Skewness . . . . . . . . . . . . . . . . . . . . . . . . 15

1.5.2 kurtosis . . . . . . . . . . . . . . . . . . . . . . . . . 17

1
❖CHAPTER 1. DESCRIPTIVE STATISTICS Dr. HAFIRASSOU Zineb

Descriptive statistics is the set of scientific methods used to collect, describe and

analyze observed data

1.1 Statistical vocabulary

❶ Population: is the set of individuals or objects of the same nature on which

the study relates.

❷ Individuals: or statistical units are the elements of the population.

❸ Sample: is a subset of the population.

❹ Statistical variable: or character X is the subject under statistical study .

❺ Statistical modality: or category the different possible situations (levels) of

a statistical variable.

There are two types of statistical variables

Quantitative variables

Are the variables that can be measured, they are characterized by numerical

values. Variables whose modalities are numbers.

A quantitative statistical variable can be:

continuous: when it can take numbers from an interval of real numbers

(measurement results).

Discrete: if it takes isolated values.

Temporal: These are particular quantitative variables that use units of mea-

surement of time. There are two types, date type (date of birth: 04/26/1994)

and time type (study hours: 6h).

2
❖CHAPTER 1. DESCRIPTIVE STATISTICS Dr. HAFIRASSOU Zineb

Example 1.1.1. .

variable possible modalities type of variable

height 1.70m, 1.60m, 1.65m, 1.75m continuous quantitative

the number of students 30, 50, 60, 80 discrete quantitative

Qualitative variables

These are variables that are not measurable (do not have numerical values).

Variables whose modalities are words.

Qualitative statistical variables can be:

Ordinal: these are variables whose modalities are ordered according to their

meaning.

Nominal: these are variables whose modalities cannot be ordered according

to their meaning.

3
❖CHAPTER 1. DESCRIPTIVE STATISTICS Dr. HAFIRASSOU Zineb

Example 1.1.2. .

variable possible modalities type of variable

eye color black, blue, green, brown nominal qualitative

degree of satisfaction very satisfied, satisfied, dissatisfied ordinal qualitative

with one’s standard of

living

❻ Statistical series: The simplest form of presenting statistical data relating to

a single character or variable consists of a simple enumeration of the values

taken by the character.

❼ Absolute frequency ni : is the number of statistical elements relating to a

given modality.

❽ cumulative absolute frequency nci ↑: the number of individuals which

correspond to the same modality and to the previous modality.

ni
❾ Relative frequency fi : the ratio n
.

nci ↑
❿ cumulative relative frequency fic ↑: the ratio n
.

Example 1.1.3. The marks of 9 students in a group are as follows

Notes ni nci ↑ fi fic ↑

5 2 2 2/9 2/9

6 1 3 1/9 1/3

8 3 6 1/3 2/3

12 2 8 2/9 8/9

16 1 9 1/9 1
5
X
Total n=9 fi = 1
i=1

4
❖CHAPTER 1. DESCRIPTIVE STATISTICS Dr. HAFIRASSOU Zineb

➀ Class (Interval): we call class a grouping of values of a variable according to

intervals which can be equal or unequal. It is mainly used when the variable

studied is continuous quantitative.

For each class we can define:

- A lower limit

- An upper limit

- Amplitude = upper limit - lower limit

lower limit + upper limit

- Class center ci = .
2

Example 1.1.4. : The blood glucose level (glycemia) in 14 subjects in g/l

class ci ni nci ↑ fi fic ↑

[0,85 ; 0,91[ 0,88 3 3 3/14 3/14

[0,91 ; 0,97[ 0,94 5 8 5/14 4/7

[0,97 ; 1,03[ 1 3 11 3/14 11/14

[1,03 ; 1,09[ 1,06 2 13 1/7 13/14

[1,09 ; 1,15[ 1,12 1 14 1/14 1

X 5
Total n=14 fi = 1
i=1

1.2 Data description

Depending on the type of variable studied. There are two forms of presentation

to describe a series of statistical data: tables and graphical representations.

1.2.1 Tables

The table can be used whatever the nature of the data, it is used to present the

data in an accurate and complete manner.

5
❖CHAPTER 1. DESCRIPTIVE STATISTICS Dr. HAFIRASSOU Zineb

1.2.2 Graphics

The objective of the graphs is to bring out a systematic vision of the phenomenon

studied by illustrating a general trend and giving an overall picture of the results.

Histogram

Histograms are surfaces that allow the representation of a continuous quantitative

variable.

Bar graphs

A bar graphs is a graphic representation reserved mainly for a qualitative variable

using rectangles of the same width.

6
❖CHAPTER 1. DESCRIPTIVE STATISTICS Dr. HAFIRASSOU Zineb

Circle graph or the Pie chart

We draw on a disk sections corresponding to the modalities of the character whose

angles are proportional to the percentages.

ni
αi = 3600 ∗ fi = 3600 ∗
n

7
❖CHAPTER 1. DESCRIPTIVE STATISTICS Dr. HAFIRASSOU Zineb

1.3 Position parameters

Central tendency or position parameters: values located in the center of the

statistical distribution which are the mean, mode and median.

1.3.1 Mean

Case of a discrete statistical variable

Let X be a discrete statistical variable and x1 , x2 , ..., xk its values for which corre-
k
X
spond the numbers n1 , n2 , ..., nk , with n = ni
i=1

k k
1X X
x̄ = ni xi = fi xi .
n i=1 i=1

Example 1.3.1. .

xi 0 1 2 3 4

ni 2 3 1 1 1
5
1X 1  12
x̄ = ni xi = 0 × 2 + 1 × 3 + 2 × 1 + 3 × 1 + 4 × 1 = = 1.5.
n i=1 8 8

Case of a continuous statistical variable

Observations are grouped into classes, so

k k
1X X
x̄ = ni ci = fi ci .
n i=1 i=1

Example 1.3.2. .

class ci ni

[1,2[ 1.5 3

[2,3[ 2.5 1

[3,4[ 3.5 2

8
❖CHAPTER 1. DESCRIPTIVE STATISTICS Dr. HAFIRASSOU Zineb

3
1X 1  14
x̄ = ni ci = 3 × 1.5 + 1 × 2.5 + 2 × 3.5 = = 2.33.
n i=1 6 6

1.3.2 Mode

Case of a discrete statistical variable

The mode Mo is the most commonly occurring value.

Example 1.3.3. .

xi 2 3 5 6 7 8 9 10

ni 2 1 1 2 2 1 1 1

Mo = 2, 6, 7

Case of a continuous statistical variable

In this case the mode is calculated by the formula

d1
 
Mo = Li + a
d1 + d2

• Li : the lower limit of the modal class (the class that has the highest frequency)

• d1 = the absolute frequency of the modal class- the absolute frequency of the

previous class (ni − ni−1 ).

• d2 =the absolute frequency of the modal class- the absolute frequency of the next

class (ni − ni+1 ).

• a: the amplitude of the modal class.

Example 1.3.4. .
class ni

[1,60-1,65[ 3

[1,65-1,70[ 8

[1,70-1,75[ 2

9
❖CHAPTER 1. DESCRIPTIVE STATISTICS Dr. HAFIRASSOU Zineb

• The modal class is: [1, 65 − 1, 70[.

• Li = 1, 65.

• d1 = 8 − 3 = 5.

• d2 = 8 − 2 = 6.
5
 
• a = 1, 70 − 1, 65 = 0.05 then Mo = 1, 65 + 0.05 = 1, 67
5+6

1.3.3 Median

Case of a discrete statistical variable

The median Me is the value at the center of a series of numbers arranged in

ascending order.

• If n is even, then
x n2 + x n2 +1
Me =
2

• If n is odd, then

Me = x n+1
2

Example 1.3.5. The number of children of 6 families is as follows

7, 3, 1, 1, 5, 2

We first order the values:

1, 1, 2, 3, 5, 7
| {z } | {z }
3 3

x3 + x4 2 + 3
We have n = 6 is even so Me = = = 2.5.
2 2

Example 1.3.6. The number of children of 7 families is as follows

3, 2, 1, 0, 0, 1, 2

10
❖CHAPTER 1. DESCRIPTIVE STATISTICS Dr. HAFIRASSOU Zineb

We first order the values:

0, 0, 1, 1 , 2, 2, 3
| {z } |{z} | {z }
3 Me=x4 =1 3

We have n = 7 is odd so Me = x4 = 1.

Case of a continuous statistical variable

In this case the median is given by

<Me
n X
− ni
2 i=1 !
Me = Li + a
nMe

• Li : the lower limit of the median class

<Me
X
• ni = the sum of the absolute frequencies corresponding to all classes below
i=1
the median class.

• nMe = the absolute frequency of the median class.

• a: the amplitude of the median class.

Example 1.3.7. According to the example (1.1.4), we obtain

• The median class is: [0.91 − 0.97[.

• Li = 0.91.

• n = 14.
<Me
X
• ni = 3
i=1
• nMe = 5 .

• a = 0.97 − 0.91 = 0.06

7−3
 
then Me = 0.91 + 0.06 = 0.958
5

11
❖CHAPTER 1. DESCRIPTIVE STATISTICS Dr. HAFIRASSOU Zineb

1.3.4 Quartiles

Case of a discrete statistical variable

Quartiles are the three values that divide the distribution into four equal parts.

• The first quartile Q1 represents 25% of the sample i.e. Q1 is the value xi whose
n
position is the smallest integer following .
4
• The second quartile Q2 represents 50% of the sample.

• The third quartile Q3 represents 75% of the sample i.e. Q3 is the value xi whose
3n
position is the smallest integer following .
4

Interquartile range

The interquartile range is the difference between the third and first quartile:

IQ = Q3 − Q1

12
❖CHAPTER 1. DESCRIPTIVE STATISTICS Dr. HAFIRASSOU Zineb

Boxplot

Example 1.3.8. In the example of the following observations

xi 1 3 5 7 9

ni 1 2 1 2 2

nci 1 3 4 6 8
n
• We have n = 8 and = 2 so Q1 is the second value Q1 = x2 = 3.
4

3n
• We have n = 8 and = 6 so Q3 is the sixth value Q3 = x6 = 7.
4

1.4 Dispersion parameters

Dispersion parameters are the parameters that summarize the dispersion of values

around the central value

1.4.1 Range

The difference between the largest value and the smallest value observed is called

the range e.

e = xmax − xmin

13
❖CHAPTER 1. DESCRIPTIVE STATISTICS Dr. HAFIRASSOU Zineb

Example 1.4.1. The marks of 10 students are as follows

2, 3, 10, 10, 11, 12, 15, 18, 19, 20

then

e = xmax − xmin = 20 − 2 = 18

1.4.2 Variance

A variance is the arithmetic mean of the squares of the differences between the

values of a variable and the arithmetic mean.

k k
1X 1X 2
V(X) = ni (xi − x̄) =
2
ni xi − x̄2
n i=1 n i=1
k
X k
X
= fi (xi − x̄) . =
2
fi x2i − x̄2 .
i=1 i=1

1.4.3 Standard deviation

We call standard deviation denoted σX the square root of the variance.

p
σX = V(X)

1.4.4 Coefficient of variation

The coefficient of variation, CV, is defined by

σX
CV =
x̄

Example 1.4.2. .

14
❖CHAPTER 1. DESCRIPTIVE STATISTICS Dr. HAFIRASSOU Zineb

xi 0 1 2 3 4

ni 2 3 1 1 1

x̄ = 1.5

k
1X 2
V(X) = ni xi − x̄2
n i=1
5
1X 2
= ni xi − (1.5)2
8 i=1
1 
= 2 × 02 + 3 × 12 + 1 × 22 + 1 × 32 + 1 × 42 − 2.25
8
32
= − 2.25
8
= 1.75

The standard deviation

p √
σX = V(X) = 1.75 = 1.3

and the coefficient of variation

σX 1.3
CV = = = 0.87
x̄ 1.5

1.5 Shape parameter

1.5.1 Skewness

There are several coefficients, the main ones are as follows:

• Pearson’s skewness coefficient

x̄ − Mo
AP = .
σX

15
❖CHAPTER 1. DESCRIPTIVE STATISTICS Dr. HAFIRASSOU Zineb

• Yule’s skewness coefficient:

Q1 + Q3 − 2Q2
AY = .
Q3 − Q1

Remark

• A zero coefficient indicates that the distribution is symmetrical.

• A positive coefficient indicates a right-skewed distribution.

• A negative coefficient indicates a left-skewed distribution.

16
❖CHAPTER 1. DESCRIPTIVE STATISTICS Dr. HAFIRASSOU Zineb

1.5.2 kurtosis

• Pearson’s kurtosis coefficient:

m4
APP =
σ4X

where m4 is the centred moment of order 4 defined by

k
1X
m4 = ni (xi − x̄)4
n i=1

• Fisher’s kurtosis coefficient:

m4
APF = −3
σ4X

Remark

• If APF = 0 then the distribution is called "normal" or "mesokurtic".

• If APF < 0 then the distribution is called "platykurtic".

• If APF > 0 then the distribution is called "leptokurtic".

17