Phase 2 : Définir et poser le

problem
The histogram

Vertical bar chart

Introduction
It is a vertical bar chart and the most graph
Current can represent the statistical distribution of a series of data or measurements concerning a
same size.
It is also used to analyze the characteristics of a population (mean value, dispersion or
spreading, behavior, trends, etc.).
The histogram is a third necessary tool to get a broader view of a problem.

A. Usefulness of the tool

It consists of representing columns (rectangles) of equal widths. The surface of the column will be
proportional to the corresponding frequency.

Frequency
(Number of rooms
defective) 18

14
12
10

6
5
3 3

A B C D E F G H Types of defects
Interval

Figure 11: Histogram of defect type distribution

Note that the histogram requires that the characteristic to be controlled is measurable.

1
B. Types of characteristics to monitor and data
The type of control that we want to implement depends on the characteristics that we want to study; we
can divide them into two classes:
 Measurable characteristics
 Non-measurable but countable characteristics
Data associated with measurable characteristics are said to be continuous type, such as, for example,
diameter, weight, lifespan, turnover, determined, viscosity,... These characteristics can
taking values in a finite or infinite interval, these are continuous variables.
Data associated with countable characteristics are said to be of discrete type, such as
exemple, nombre d'unités non conformes, nombre de non-conformités, pourcentage de non conformes,...
These characteristics take only a limited number of values, often integer values; they are
discrete variables.
In the following sections, we will discuss some descriptive data analysis techniques.
who are the most used in the industrial environment. We indicate how to characterize a feature
of quality by certain important statistics and by certain appropriate figures and graphs.

C. Data processing: distribution of absolute frequencies and histogram

Starting from a dataset associated with an important characteristic of a process, we want to be
Able to provide a concise and intelligible presentation of the feature values.
A simple way to summarize a data series is to arrange them in ascending order (this step is not
not essential, but can facilitate the processing) to process them according to a frequency distribution
Absolutely visualize this counting using a histogram.

This initial analysis allows for the localization of the center of the distribution of the values of the characteristic.
as well as to get a good idea of the dispersion (variation) of the values.
Although the software available on the market allows for a quick distribution of
absolute frequencies as well as the plotting of the histogram, we still illustrate the procedure to
follow since it frequently happens that in the factory, the person responsible for the adjustment of certain
features to carry out this counting himself and make the necessary corrections if applicable.

1) Tallying and distribution of absolute frequencies

Grouping of data into classes where each data point is indicated by a vertical line.
belonging to its respective class is called data stripping. It is also common practice
current practice of processing the data in blocks of 5 (if applicable). The sum of the number of features belonging
for each class, give the absolute frequency of that class (which corresponds to the number of data)
belonging to this class). The distribution of data in the classes along with the frequencies
absolute frequencies are called the distribution of absolute frequencies or the distribution of counts.
When you want to group a series of data according to a distribution of absolute frequencies, you must
first, set the number of classes into which the values are distributed. A bit of experience and
The following tips can make the task easier.

Determination of the number of classes

Let us first mention that the number of classes should generally be neither less than 5 nor greater than
Preferably, it will range between 6 and 12 classes.
This choice depends on the amount of data to be processed and the scattering of the data.

2
We could use the table opposite to establish the number of classes.

Data name n Number of classes k

< 50 5 to 7
50 to 100 6 to 10
100 to 250 7 to 12
> 250 10 to 20
Table 1: Table for Determining the Number of Classes
(Ishikawa Table)(*)

Kaoru Ishikawa, "Quality Management Tools and Practical Applications" Dunod Edition 1990

Note:
The number of classes can be determined from the following relationship: k = n 1/2
One must also round up the value in order to be able to allocate all the observations.
data" to all corresponding classes.
3) Determination of the amplitude of each class
Here to avoid any confusion in the presentation of results as well as in the representations
graphs that can follow, we will ensure as far as possible that each class is represented
with the same amplitude.
To find the amplitude of the classes, we can proceed as follows:
Note, using the previous table, the desired number of classes according to the number of
1
data to be processed.
Determine the largest (xmax) and the smallest value (xminof the series and calculate afterwards
2
the extent of the series as follows: E = xmin-xmax.
We then divide the range of the series by the desired number of classes: E/k. This gives us a
idea of the amplitude that each class should have. Since this result will rarely be a
3 whole number, we round it up to the nearest greater or smaller integer. The final choice of
the amplitude of each class will be carried out in order to ensure the maximum clarity possible and to
facilitate the presentation and understanding of the frequency distribution.

We can visualize the frequency distribution using the histogram.

4) Histogram
The histogram is a graphic representation of the frequency distribution and is made up of
juxtaposed rectangles whose bases are equal to the interval of each class and whose height
is such that the surface is proportional to the absolute or relative frequency of the corresponding class.
In the case where the classes have the same amplitude, height = frequency.

Remarks

We obtain the relative frequency (in %) of each class by dividing the absolute frequency or count.
of each class by n, the sample size:
fi% = (fi/n) x 100.

3
Or well fI% = (ni(n) x 100.

In the case of unequal class amplitude, the heights must be adjusted as follows so that the surface
of each rectangle is proportional to the frequency:
If the amplitude of an absolute frequency class is m times greater (or smaller) than the amplitude
Initially, its rectangle will have a height of f/m (or m . f).
Thus, in the following distribution:
Classes Frequencies
10–20 5
20–25 4
25–30 8
...

The height of the histogram for the first class will be 5/2 = 2.5 since the amplitude
of this class is double (amplitude = 10) compared to the other classes (amplitude = 5).

5) Example: Weighing in grams of a sealing point

The following data represents the weight in grams of a sealing joint obtained from a process.
continues and used in the manufacturing of automobiles, each value corresponds to a period of
30 seconds of manufacturing. The variation in the flow of rubber from the extruder
directly affects the dimensions of the sealing gasket. Forty data points were obtained on a
production period of about 30 minutes and they represent the sample size (n = 40).

Data obtained under standard operating conditions

269.7 267.0 264.8 267.1 268.7
263.6 265.6 261.4 265.5 261.2
264.4 268.8 264,5 264.5 263.1
259.7 260.3 266.2 262.2 264.6
262.4 263.4 265.9 271.0 258.7
263.4 267.6 265.3 264.4 262.3
260.7 264.1 266.4 269.8 261.2
265.0 272.9 255.8 266.1 262.1
Table 2: Weight in grams

We want to establish the distribution of absolute frequencies of weight in grams.

1 Since n = 40, the desired number of classes is k = 6.
The greatest value of e is: xmin= 255.8
2
The extent of the series is therefore: E= 272,9 - 255,8 = 17,1.
The amplitude of each class must be equal to E/k = 17.1/6 = 2.85 ≈ 3
As the smallest value of the series is 255.8, one could set the lower limit of the
first class at 255.5 with an amplitude of 3 grams for each class.
3 We then obtain the following classes:
255.5 ≤ Weight < 258.5
258.5 ≤ Weight < 261.5
261.5 ≤ Weight < 264.5

4
 264.5 ≤ Weight < 267.5
267.5 ≤ Weight < 270.5
270.5 ≤ Weight < 273.5
In this way, the upper limit is never included in its class; all classes
are mutually exclusive, each data belonging to a single class in the
Tabulation. We could use the following sheet to facilitate the tabulation.

Counting sheet
Frequency distribution
Absolutes
Characteristic: Weight in Part no.:……………………………
grams Joint - rubber
Specifications:…………… Process: Extrusion
Sample size: 40 Department:
Date: 12/05 Extrusion 742
Frequency
Effective Frequency
Classes Counting cumulative
ni (fI%)
(FcI↑%)
255.5 - 258.5 / 1 2.5 2.5
258.5 - 261.5 //// // 7 17.5 20.0
261.5 - 264.5 //// //// / 11 27.5 47.5
264.5 - 267.5 //// //// //// 14 35.0 82.5
267.5 - 270.5 //// 5 12.5 95.0
270.5 - 273.5 // 2 5.0 100.0
Total n=40 100%
Table 3: The seal point (weight measurement in grams)

The cumulative frequency Fc↑% corresponds, in our case, to the percentage of data in the series that are
less than the upper limit of the class. Since the range of each class is the same, we
obtains the following histogram, using absolute frequencies on the y-axis and the intervals
of classes on the x-axis.
The sealing point (measurement of the poité) chosen according to the table for determining the number of classes or table
of Ishikawa

6) Histogram of weight in grams

18
16
14
Effective 12
Frequency 10
absolute)
8
6
 Figure 12: Histogram of weight in grams

The central value of a class or midpoint is obtained by adding, to the lower limit of the class, the
half of the amplitude of the class. Here we obtain 255.5 + 3/2 = 257 for the first class; the others
class centers are 260 263 266 269 272. We could also present the histogram by indicating
the centers of the class instead of the limits of each class.

7) Histogram with class centers

18
16
14
Effective
(Frequency 12
absolute 10
8
6
4
2
0
257 260 263 266 269

Figure 13: Histogram of weight in grams with class centers

It is observed that a significant percentage of the data is between 261.5 and 267, the central value of the
weight around 264 g, we also observe the spread of the data between 255 and 273 g.
with no gaps in the distribution or with no extreme values.
Remarks

1) In the drawing of a histogram, it is advised not to confuse the y-axis in the drawing.
(absolute or relative frequencies) and the first rectangle of the histogram. Therefore, it is necessary to separate
the origin of the abscissas from the origin of the ordinates.

2) It should also be specified that the way of representing a series of data (continuous variable) at
The aid of a distribution of absolute frequencies and a histogram accepts as a hypothesis
simplifying, the uniform distribution of data within each class. Ultimately, the

6
Values grouped in the same class will be assigned the same value, namely that of the center of the
class.
3) If you want to compare histograms made from samples of different sizes, it will be
It is therefore preferable to use relative frequencies instead of absolute frequencies on the y-axis.

D. Use of the histogram to describe and control a process

The histogram is a relatively effective statistical tool to provide an 'image' of behavior.
an industrial process. Because of its simplicity, it has the advantage of being easily understood, even for
those unfamiliar with the concepts of statistics.
The operator of a machine for which he is responsible for the settings makes a sensational discovery when
can visually represent the behavior of a certain characteristic corresponding to the
part that he manufactures.

E. What information can be obtained from a histogram?

The histogram allows you to see at a glance where the concentration of values is.
observed, the extremes that occur more rarely, the overall shape of the distribution of
data,...
In particular, it allows for a fairly accurate indication of the main characteristics of the process.
of manufacturing,
 Indicating the Center of the Process;
 By indicating the spread of the data, which provides a good idea of the dispersion of
values (important aspect that allows for evaluating the accuracy of the manufacturing process by
example).

F. Utilité de l'histogramme en industrie

The use of the histogram (or the distribution of absolute frequencies) in the industrial field
can therefore serve:
 To check the setting of a machine.
 To indicate the corrective measures to be taken at the beginning of production.
 To measure the effects of corrective measures.
 To assess whether a machine can produce in compliance with the specified standards for the
characteristic that we control.
 To determine if there is a mix of raw materials with different properties.
 To compare different operators assigned to the same process.
 To compare different suppliers (receiving control).

Note. To obtain a more comprehensive assessment of the main values of the process
(center and dispersion), we must calculate certain descriptive measures (such as the mean and the standard deviation)
These concepts are addressed in the following sections.

G. Measures of central tendency and dispersion

Although the histogram (or the stem-and-leaf plot) provides us with valuable information about a
quality characteristic, it does not allow for precise specification of the central value of the
characteristic as well as the dispersion.
To complete the characterization of a statistical variable, we mainly use two types of
measures:
 Measures of central tendency

7
  Measures of dispersion (or variability)
Measures of central tendency provide an accurate idea of the order of magnitude of the values.
as well as the central value of the quality characteristic.
Dispersion measures, on the other hand, allow us to quantify the scattering of the values of the
characteristic and to specify the extent to which the observed values deviate from each other
others or deviate from their central value.
The most commonly used measurements in industrial environments are summarized in the table below.

Average
Trend arithmetic
central Median
Mode

Etendue
Standard deviation
Dispersion
Coefficient of
variation

Figure 14: Measurements used in industrial environments

H. Definitions of measures of central tendency

Arithmetic mean
The arithmetic mean, which we denoteXof a ̅ numerical series x1, x2, x3, x4, is the sum of
values of the series divided by the number n:

1+ 2+ 3+ …….... ∑ni=1
̅X = =
n
̅ read as "x bar"∑(grand sigma) designates the sum of.
The symbolXis

∑ =1 . ∑ =1 .
̅X = =
∑ =1 1

If the characteristic takes distinct values xi(or even if the values are grouped into classes with
xias class centers), with a certain number of repetitions niabsolute frequency of value xi
or from the class where the center x is locatediyou ci), we then use the following expression:
k representing the number of distinct values in the series or the number of classes in the series
grouped.

Median
The median, denoted as Meor X, is the value (observed or possible) of the characteristic in the series
of data arranged in ascending or descending order, which divides this series into two parts, each
having the same number of data points on both sides of the median.

Odd number of data (not grouped into classes): the median then corresponds to (n+1)ème2 value
from the ordered series. There are therefore (n-1)eme2 values on each side of the median.

8
Even number of data (not grouped into classes): In this case, the median will be the average
arithmetic of the two central values in the ordered series. Thus if n = 2k, the median is the
average of ke(k+1)evalues.

In the case where the data is grouped into classes, one can obtain the median by performing a
linear interpolation within the median class; we then use the following expression or method:
Interpretation of the method

Me = BI + a . [ (n 2̸ )–F ] f̸ Me
BI Lower bound of the middle class
N The total number of data in the series
The sum of the absolute frequencies of all
F
the classes preceding the median class
fMe The absolute frequency of the median class
A The amplitude of the middle class
Table 4: Interpretation of the method

To determine the median class, it is necessary to determine the quantity n/2 (which corresponds to 50% of the
data) and we compare this value with the cumulative frequencies. The median class will be the one whose
th
cumulative frequency encompasses n/2 data (the one whose cumulative frequency is immediately
greater than or equal but not less than.

3) Mode
The mode (or dominant value), denoted Mois the value of the most frequent variable that we observe
in a series. In the case of a discrete variable, determining the mode is immediate. In the case
of a continuous variable whose data has been grouped into classes, determining the mode is little
objective and is rather left to arbitrariness. In this case, we refer more to modal class, the class to which
corresponds to the highest frequency (absolute or relative). By convention, one could say that the
The mode is then the value that corresponds to the modal class center.

4) Definition of measures of dispersion

a) Scope
The range is the difference between the largest and smallest values in the series (of a sample or
of a subgroup E = XmaxXmin

This measure only takes into account the extreme values of the series; it is also noted that it is
independent of the number of data in the series. However, it is little used when the number of
data is 10 and more. This measure of dispersion is frequently used in monitoring and control
statistics of processes (part 3).

b) Variance and standard deviation

The dispersion of the values xIof the series around their average x is obtained by calculating the sum of the
squared deviations of the values xiwith respect to x, divided by (n-1). This measure is called the variance of
the series of values (or of the sample) is written as:

2=
∑ =1( - ̅X)2 ∑ =1
2
−( ∑ ) 2
⁄
=
−1 n− 1

9
The square root of s2give the standard deviation: S= √ 2

The arithmetic mean and the standard deviation are expressed in the same unit of measure as that of the values.
xiof the observed characteristic. A series that is little dispersed around the arithmetic mean (that
which is desirable in industrial control) leads to a low standard deviation value.
In the case of distinct values xior values grouped into classes with xias class centers and
absolute frequency fIwe use the following expression for calculating variance:

2
2=
∑ − (∑ )2 ⁄
−1

c) Coefficient of variation

The coefficient of variation, which we denote as CV, is obtained by dividing the standard deviation by the mean.
Arithmetic x bar. Expressed as a percentage, it is written as:

= S̅ × 100, X≠
̅ 0

It is independent of the unit of measurement of the observed characteristic. If X is negative, we will remember
So the absolute value of CV. The lower the coefficient of variation, the more the data series is
homogeneous (concentrated aroundX indicating thus that the average X is well representative of
the entire dataset of the series.

10
 Speed V 1
eleven
0.148 0.147 0.147 0.145 ......... 0,145 1,756
Average diameter X 0.14633
12 12 12
The extent E 0.148 0.145 0.003
xié   2
xi /n 0.256976  1,756  2/ 1 2
Variance s 2 0.0000013
11
The gap type 0.0000013 0.00114
s 0.00114
Coefficient of variation CV % C100 100 0.77 9%  less than 1% .
x 0.14633

Speed V 2
1,789
X 0.1490833
12
E 0.151 0.147 0.004
2

0.266727 1,789  2/12 0.0000169
s 0.0000015
11 11
s 0.0000015 0.00124
0.00124
CV% 100 0.832%
0.1490833

Speed V3
We can verify that we obtain the following descriptive measurements:
X 0.15292, E 0.004 s2 0.0000013 s 0,00114,CV% 0.761%..
I. Synthesis of descriptive measures according to the speeds of the extruder
(Outer diameter of the tubes)
V1 V2 V3
Average 0,14633 0.1490833 0.15292
Extent 0.003 0.004 0.004
Standard deviation 0.00114 0.00124 0.00114
CV % 0.789% 0.832% 0.761%
Table 5: Outer diameter of tubes

11
J. Exercises
Exercice : 1
In a manufacturing process of thermoplastic tubes, the following values were obtained for the
external diameter of the tubes according to three manufacturing speeds, under the influence of various factors
production :

V1 V2 V3
0.148 0.148 0.148 0.148 0.148 0.148 0.148 0.148 0.153 0.152 0.152 0.153

0.145 0.145 0.145 0.145 0.149 0.149 0.149 0,149 ["0.154","0.154","0.152","0.152"]

0.147 0.147 0.147 0.147 0.149 0.149 0.149 0.149 0,153 0,151 0,154 0,155
Table 6: Data on Manufacturing Speeds

Calculate the arithmetic mean, the range, the variance, the standard deviation, and the coefficient of variation for
the outer diameter of the tubes according to each speed of the extruder.
Solution
Exercise: 2

Calculation of the mean and variance - grouped data

The following data represents the starting voltage in volts of a fluorescent lamp; forty
values have been obtained.

Tension
0 91 92 93 94 95 96 97 98 99
(xi)
Number of lamps:
01 02 02 12 07 09 04 00 02 01 40
Effective nI
Frequency: fi
Table 7: Data concerning the starting tension
in volts of a fluorescent lamp

a) Determine the mean and variance using the expressions for grouped values:

∑ =1 . ∑ =1 .
̅X = =
∑ =1 1
:
2
2=
∑ − (∑ )2 ⁄
−1

12
To facilitate the calculation, use the following table:

xi xI 2 ni fI fixi fixi 2
0 01 0.025
91 02 0.05
92 02 0.05
93 12 0.3
94 07 0.175
95 09 0.225
96 04 0.1
97 00 00
98 02 0.05
99 01 0.025
01 0.025

Table 8: Determination of the mean and variance

a)X̅ = … … …. , S2=…………..
b) Determine the coefficient of variation (in %). CV =
c) What is the modal value of this data series?

Solution
a) X =93.35 ;s23.8774
b) CV= 2,11 %
c)92

Exercise: 3

Determination of the median lifespan of a product

We subjected 12 identical products to a reliability test, for which we found the lifespans.
following in hours.

76 157 116 85 120 285 211 159 184 138 92 101

As the lifespan varies widely, the arithmetic mean will not be very representative.
Let's determine the median value instead.

Let's arrange the series in ascending order:

76 85 92 101 116 120 138 157 159 184 211 285

Since the number of observations is even, n = 12 = 2k, the median is the average of the 6th and 7th.
observation in the ordered series.

Me [120 + 138] ̸ = 129 hours. We can therefore say that there are six products that have a lifespan.
less than 129 hours and 6 that had a lifespan greater than 129 hours.

13
Exercise: 4

Determination of the median value of a grouped series

The weight distribution (X) in grams of fifty ceramic tubes is summarized in the
the following absolute frequency distribution. Complete the table by adding the cumulative frequencies
croissants and determine the median value of the weight.

Weight in grams Name Of cumulative frequencies

Less than 1.45 5
1.45 < X < 1.60 8
1.60 < X < 1.75 10
1.75 < X < 1.90 15
1.90 < X < 2.05 9
2.05 < X < 2.20 2
2.20 and more 1
Table 9: Distribution of the absolute frequencies of 50 ceramic tubes

What is the median class?

b) Indicate what the values are for the following quantities:
BI=……………., n=……………., F =……………. , f M e% =………………., a =…………….
c)______________________Calculate the median using the following expression:

Me = BI + a . [ (n 2̸ )–F ] f̸ Me

a) Solution

Weight in grams Cumulative frequencies

Less than 1.45 5
1.45 ≤ X < 1.60 8
1.60 ≤ X < 1.75 10
1.75 ≤ X < 1.90 15
1.90 ≤ X < 2.05 9
2.05 ≤ X < 2.20 2
2.20 and more 1
Table 10: Distribution of cumulative increasing frequencies

The median class is 1.75 ≤ X < 1.90.

ii) BI=1,75, n=50, F =23, f Me% =15, a =0,15
iii) Me1.77g

Remarks.

a) The expression of the median for grouped values in classes assumes that the values in the
the median class and are uniformly distributed.
b) We cannot calculate the arithmetic mean in the case of an open class distribution like
this is the case for exercise 1.5, we must then use the median as a measure of central tendency.
c) The median, unlike the arithmetic mean, is not influenced by extreme values.
éventuellement très grandes ou très petites. Elle est toutefois influencée par le nombre de données.

14
d) If the characteristic is discontinuous, there may be no median value. The median must
correspond to a possible value of the characteristic.

e) A distribution is symmetric if the data of the characteristic are equally dispersed on both sides.
and others with a central value. In the case of a symmetric distribution, we have
mean = median = mode. If the skewness is positive, then X MeMo, if the asymmetry is negative, X
< Me< Mo.

Exercise: 5

The metal company manufactures metal rods used across the country by various clients in
the assembly of certain metallic structure assemblies. An important characteristic of these rods is the
tensile strength that allows to assess the ability of the rods to withstand a certain effort.

The company's manufacturing process provides rods of good quality, but we notice
however, the tensile strength may vary slightly from rod to rod. this fluctuation is
attributable to several factors including the mechanical properties of raw materials, light grooves
in the stems,... a test conducted on thirty-five stems produced the tensile strengths
Presented in the table below, measurements taken using a static testing machine.

390 384 340 385 381

384 376 420 404 355
423 361 317 383 427
437 365 380 370 382
292 470 409 396 498
378 376 349 402 364
331 342 386 412 408
Table 11: Data concerning tensile strength (kg/cm2)

a) Identify the quality characteristic.

b) What is the unit of measurement for the characteristic in question?
Sort the data in ascending order.
d) Quelle est la résistance a la traction la plus faible? la plus elevée?
e) What is the extent of the series?
f) What is the desired number of classes for the distribution of absolute frequencies? What will it be then?
therangeofclasses?
g) Using 290 as the lower limit of the first class and 35 as the class width,
strip the data according to a distribution of absolute frequencies, noting that x represents the
tensile strength of the rods.
h) In which class is the largest number of stems located?

Exercise: 6

In a manufacturing workshop, the Rockwell hardness was measured after hardening sixty pieces.
mechanics. The data in ascending order are presented in the following table.

15
 50.9 51.4 51.9 52.5 52.9 53.0
53.5 53.6 53.8 54.0 54.0 54.2
54.4 54.6 54.6 54.6 54,7 54,7
54.9 55.0 55.0 55.1 55.1 55.2
55.2 55.2 55.2 55.4 55.5 55.5
55.6 55.6 56.0 56.0 56.1 56.2
56.2 56.3 56.4 56.5 56.6 56.7
56.7 56.9 57.0 57.1 57.1 57.2
57.4 57.6 57.8 58.0 58.1 58.3
58.9 59.0 59.3 60.0 60.2 60.6
Table 12: Data concerning the measurement of Rockwell hardness after quenching

a) Process these data according to an absolute frequency distribution using a hardness of 50.5
as the lower limit of the first class and 1.5 as the amplitude of each class.
b) Quelle est la dureté de la pièce la plus douce? quelle est la dureté de la pièce la plus dure?
c) It is specified that the parts found in the following intervals are classified as soft, medium,
hardships.
soft parts: hardness within the range 50.5 < x < 53.5
medium parts: hardness within the range 53.5 < x < 56.5
hard parts: hardness within the range 56.5 < x < 61.
Howmanypiecesareclassifiedineachofthesecategories?

Exercise : 7

The company gimtek manufactures heating plates for domestic use. however, the company
supplies control thermostats from anAmerican supplier.
A reception inspection is carried out by 10.1 11.3 12,4 12.1 13.0 12.6
a technician who checks using a 10.6 13,8 12.6 12.0 13.4 11.1
sample of thirty thermostats, the 10.8 14.1 10.6 11.3 11.5 11.9
minimum temperature at which a baseboard 11.6 9.6 11.0 12.9 12.0 12.0
heating can operate with these thermostats. 10.0 12.3 12.8 11.8 11.3 12.4
the results on the control of a recent batch Table 13: Data regarding temperature in °C
received by the company are presented in the
table above.
Does this control pertain to a measurable or countable quantity?
b) Ordonner les valeurs observées (ordre croissant).
c) What is the temperature range of this sample?
d) Strip these data using 9.6 as the lower limit of the first class and 0.8 as
amplitudeofeachclass.
e) Draw the histogram.
f) The company gimtek applies the following decision rule in the control of thermostats:
 if 1 thermostat in the sample only operates at a minimum temperature of 14°c or higher, the batch is
completely verified.
 if 2 or more thermostats in the sample operate only at a minimum temperature of 14°c or
Moreover, the batch was returned to the supplier without further verification.
According to the results of the inspection carried out by the technician, what action should we consider?

K. Case study: Use of the histogram for machine adjustment

16
The company 'Luminor' manufactures incandescent bulbs. To do this, the machine M12 emits a
certain quantity of special powder on the lower walls of the bulbs, to make them opaque.
The thicker the layer of powder, the more opaque the bulb is. The measurement of opacity is obtained
by measuring the deflection of a beam of light passing through the bulb. The higher the deflection,
the lower the opacity. The desired deflection is in the range [22; 28].
Mr. Ali, in charge of the M12 machine, obtains the following deflections for 40 bulbs.
(sorted sample).
23.1 24.5 24.9 25.3 25.7 26.0 26.5 26.7
23.4 24.6 25.0 25.5 25.7 26.0 26.6 27.0
23.7 24.8 25.2 25.5 25.8 26.2 26.6 27.3
24.3 24.8 25.3 25.6 25.9 26.4 26.7 28.2
24.3 24.8 25.3 25.7 25.9 26.4 26.7 28.5
Table 14: Data on deflection

a) Fill out the tally sheet for this data?

n = ............ So k = ............ and A =...............

Characteristic noted: Agent: …………………………………….

Sample size: n = …….. Place/Machine: ……………………..
Date: ……/……/2006 Observation: ...
Ci
Center ni niCI nICi2 fi% fCi = Fi
Classes
class
No Interval

ni= n niCi niCI2 fi

Total

Table 15: Example of a structured data processing form

a) Plot the histogram of absolute and relative frequencies?

b) Combien y-a-t-il de valeurs sous 22 ? Combien y-a-t-il de valeurs sur 28 ?
c) What is the value of the average? interpret?
d) What is the value of the median? Interpret?
e) What is the value of the mode? Interpret?
What do you deduce about the shape of the measured deflection distribution?
g) What is the range of the data? the variance and the standard deviation? interpret?
h) What should the average be to have a respected interval [22; 28]?
i) Combien de valeurs sont inférieures à 25 ? Combien de valeurs sont supérieures à 25 ?

17
The production manager tells Mr. Ali that machine M12 is producing more powder than it should.
What do you think?
What would you advise Mr. Ali?

Number of classes (or intervals) k?

First, let us mention that the number of classes should generally be neither less than 5 nor more than
Preferably, it will vary between 6 and 12 classes.
This choice depends on the amount of data to be processed and the dispersion of the data.

L. Conclusion of phase 2

This phase leads to a definition of the core problem, documented and validated. It allows for
ensure the following key points:
 Have we collected and analyzed the relevant data?
 Have we focused on the heart of the problem?
 Is this vision shared within the organization?

18