D.J.

DUNN 1
MATHEMATICS FOR ENGINEERING

STATISTICS

TUTORIAL 1 BASICS OF STATISTICAL DATA

This tutorial is useful to anyone studying engineering. It uses the principle of learning by
example.


On completion of this tutorial you should be able to do the following.


Explain the use of raw data.

Present data as frequency polygons, bar graphs and histograms.

Explain and find the mean and median values.

Explain and plot ogives.

Explain and find quartiles.

Explain and find the standard deviation and variance for grouped and ungrouped
data.


D.J.DUNN 2
1. INTRODUCTION

Statistics are used to help us analyse and understand the performance and trends in various areas of work.
These might be financial trends, things to do with the population or things to do with manufacturing.
Often we wish to present information visually with easily understood graphics and so a variety of graphs
and charts are used for this purpose. Here is an example of a PIE CHART and a BAR CHART showing
the number of parliamentary seats won by main political parties at the 2001 British general election.

The number of MPs elected must be a round or whole number so the raw data is exact. When presenting
other forms of data this is not the case as explained in the following section.


2. RAW DATA

Lets use an example to get started. Consider a set of statistics compiled for the height of children of age
10. First we would compile a table of heights. This would be the raw data. Note that the larger the sample
we take, the more meaningful the results will be. Suppose we measure the heights to an accuracy of 0.01
m. This is the table of raw data for 10 year old children

Sample 1 2 3 4 5 6 7 8 9 10 11 12
Height (m) 1.45 1.56 1.37 1.44 1.32 1.42 1.55 1.29 1.37 1.49 1.47 1.34
Sample 13 14 15 16 17
Height (m) 1.56 1.28 1.35 1.62 1.46

3. RANKED DATA

If the raw data is rearranged from shortest to tallest we have the ranked data.

Sample
number
1 2 3 4 5 6 7 8 9 10 11 12 13
Height
(m)
1.28 1.29 1.32 1.34 1.35 1.37

1.37 1.44 1.44 1.46 1.46 1.47 1.49

Sample
number
14 15 16 17 Total
Height
(m)
1.55 1.56 1.56 1.62 24.34

Data presented in this form is also called DISCRETE DATA because we jump from one value to another in steps,
in this case steps of 0.01 m.

D.J.DUNN 3
4. CLASS or BANDS

If we used exact measurements of heights, it is unlikely we would find two children exactly the same height so we
round off the values. This causes problems as we shall see. When handling large numbers of samples, we end up
with huge lists of data so to simplify the table we create bands or classes within which the rounded measurements
fall. The more we round off the values, the more likely it becomes that we will find more than one in a given class.
The number of children within each class is the frequency. Next we would have to go through the laborious task of
counting how many there are in each class. If we found a child with a height exactly on the edge of the class edge,
we might decide to allocate a half to each class on either side resulting in frequency values that are not whole
numbers. The result is a FREQUENCY DISTRIBUTION TABLE.
Height 1.2 - 1.3 1.3 - 1.4 1.4 - 1.5 1.5 - 1.6 1.6 - 1.7
Mid Point or MARK 1.25 1.35 1.45 1.55 1.65
Freq. 2 5 6 3 1

5. GRAPHS

If we plot frequency vertically against height horizontally, we get a frequency distribution graph and this can be
drawn in different ways. The values plotted are the mid point values called the MARK.

These plots simply tell us the numbers in each class by the
mark. If we want to illustrate the width of the band we use a
HISTOGRAM. Notice that the mid point of each band is the
MARK. The boxes are drawn between the CLASS LIMITS.
Because the heights were rounded off to 0.01 m the limits are
0.005 either side of the CLASS BOUNDARY and the class
boundary is the exact dividing line between each class. The
width of the band from mark to mark or boundary to boundary
is the CLASS INTERVAL. In this example the interval is 0.1.

Notice that the points are drawn for the middle of the band.
D.J.DUNN 4
6. MEAN

This is one of the more common statistics you will see and it's easy to compute. All you have to do is add
up all the values in a set of data and then divide that sum by the number of values in the dataset. For our
example, let the height be represented by the variable x and the frequency be f.

Sample
number
1 2 3 4 5 6 7 8 9 10 11 12 13
Height
(m)
1.45 1.56 1.37 1.44 1.32 1.42 1.55 1.29 1.37 1.49 1.47 1.34 1.56

Sample
number
14 15 16 17 Total
Height
(m)
1.28 1.35 1.62 1.46 24.34

The mean value is denoted x and x = 24.34/17 = 1.432 m
We can do this a bit more simply using the frequency distribution table.
x (mid pt) 1.25 1.35 1.45 1.55 1.65
f. 2 5 6 3 1 total
f x 2.5 6.75 8.7 4.65 1.65 24.25

The mean value is x = 24.25/17 = 1.426 m. This is not quite as accurate as the previous answer because
the values have been taken at the mid point of the band.

7. MEDIAN
Whenever you see words like, "the average person ...", or "the average income of ... you don't always
want to know the mean. Often you want to know the about the one in the middle. That's the median.
Again, this statistic is easy to determine because the median literally is the value in the middle. In order to
find it, you just line up the values in your set of data from largest to smallest. The one in the dead-centre
is your median. Our table would look like this.

Sample
number
1 2 3 4 5 6 7 8 9 10 11 12 13
Height
(m)
1.28 1.29 1.32 1.34 1.35 1.37

1.37 1.44 1.44 1.46 1.46 1.47 1.49

Sample
number
14 15 16 17 Total
Height
(m)
1.55 1.56 1.56 1.62 24.34
The mid point in the table is point number 9 so the median value is 1.44 m. If we had an even number of
samples, say 18, then there would be two values in the middle and we should average the two to get the
median.
D.J.DUNN 5
8. MODE

The mode is the most frequently occurring value. In the example repeated below, this will be the class
with a mark of 1.45 since there are six in this group.

x (mark) 1.25 1.35 1.45 1.55 1.65
f. 2 5 6 3 1 total
f x 2.5 6.75 8.7 4.65 1.65 24.25

It is quite possible that the mode is not unique because the same maximum figure could occur more than
once the distribution. If we don't have grouped data, there is no mode unless several occur at the same
precise height. This leads us on to the next section.
9. OGIVE and QUARTILES
If we add a new row to our data showing the accumulative frequency and plot the data against it, we get a
different sort of graph called an OGIVE that makes it easier to spot the median. Lets add a new row to
our frequency distribution table containing the cumulative frequency. We can plot the same data using
bands of 0.1 m. We plot the upper limit of each band so that each frequency shows how many children
are either shorter or taller than that value. The table to plot is as follows.

The vertical scale is usually turned into % of the maximum as shown. The 50% level corresponds to the
median.

x (upper point) 1.3 1.4 1.5 1.6 1.7 Total
f. 2 5 6 3 1 n =17
f x 2.5 6.75 8.7 4.65 1.65 24.25
cum. f 2 7 13 16 17


The range that covers 75% to 100% is called the upper quartile. The range that covers 0% to 25% is
called the lower quartile. The range between 25% and 75% is called the inter quartile and this can be
divided into two parts called the semi-inter-quartile. These tell us something about how the samples are
spread around the median but a better method of doing this is to use the standard deviation. Any range
corresponding to a change of 1% is called a percentile.
D.J.DUNN 6


WORKED EXAMPLE No.1

A company manufactures steel bars of nominal diameter 20 mm and cuts them into equal lengths.
The diameter of each length is measured at the middle for the purpose of quality control. The results
for 20 bars are given below.

Produce a frequency distribution table using bands of 0.1 mm.
Calculate the mean of the samples.
Draw a histogram.
Plot the Ogive.
Determine the median, mode, upper and lower quartiles and the semi- inter-quartile.

Use the tables and plots to show your solutions

Sample 1 2 3 4 5 6 7 8 9 10
diameter 19.9 19.8 20.1 19.9 19.7 20.1 20.0 19.6 19.7 20.1

Sample 11 12 13 14 15 16 17 18 19 20
diameter 20.2 20.0 19.9 19.8 20.1 20.0 19.7 19.6 19.9 20.2

SOLUTION

Total = 398.3 Total samples n = 20 mean = 398.3/20 =19.915

RANKED ORDER

Sample 1 2 3 4 5 6 7 8 9 10
diameter 19.6 19.6 19.7 19.7 19.7 19.8 19.8 19.9 19.9 19.9

Sample 11 12 13 14 15 16 17 18 19 20
diameter 19.9 20.0 20.0 20.0 20.1 20.1 20.1 20.1 20.2 20.2

The median is the middle value of the samples when they are ranked in order and this is the 9
th

sample so the median is 19.9 mm.

The most frequently occurring values are 19.9 and 20.1 so there is no unique mode.

Total = 398.3 Total samples n = 20 mean = 398.3/20=19.915

d 19.55- 19.65- 19.75- 19.85- 19.95- 20.05- 20.15- 20.25 Totals
Mark 19.6 19.7 19.8 19.9 20 20.1 20.2
f 2 3 2 4 3 4 2 20
f d 39.2 59.1 39.6 79.6 60 80.4 40.4 398.3
cum.f 2 5 7 11 14 18 20

n = 20 Mean = 398.3/20=19.915

D.J.DUNN 7

Median =19.92
Upper quartile = 20.25 - 20.07 = 0.18 Lower quartile = 19.75 - 19.55 = 0.2
Inter-quartile range = 20.07 - 19.75 = 0.32 Semi-inter-quartile = 0.32/2 = 0.16



SELF ASSESSMENT EXERCISE No.1

The diameters of a number of components are measured to the nearest 0.1 mm. The distribution is
shown.
Diameter mm 9.6 9.7 9.8 9.9 10.0 10.1 10.2 10.3 10.4
Number 3 9 36 88 122 90 44 7 1

Draw the histogram and the Ogive and deduce the mean, the median, the upper and lower quartile
and the semi-interquartile.


D.J.DUNN 8

SOLUTIONS
The ogive looks like this

0
50
100
150
200
250
300
350
400
450
9.6 9.8 10 10.2 10.4 10.6


(mean = 10.001 mm, median = 10 upper quartile = 3.6 lower quartile = 2.68
semi-interquartile = 0.9)





D.J.DUNN 9
10. Variance and standard deviation

Standard deviation is a more difficult concept than the others we've covered. To understand this concept,
it can help to learn about what statisticians call normal distribution of data. Most statistical samples
produce a lot of results around the mean. Further away from the mean there are not so many results. The
frequency distribution graph produces a bell shaped curve called the normal distribution. However,
some are more normal than others and some times there are a lot more results close to the mean than
others.

Consider the three cases shown on the diagram.
The distribution shown by graph B has most of
the examples in the set of data close to the
"average," while those in graph A are more
widely spread from the average. In reality few
examples tend to one extreme or the other. The
standard deviation is a statistic that tells
you how tightly all the various examples are
clustered around the mean in a set of data.

When the examples are tightly bunched together
and the bell-shaped curve is steep, the standard
deviation is small (graph B). When the examples
are spread apart and the bell curve is relatively
flat, that tells you that you have a relatively large standard deviation (graph A). Lets first try to
understand graphically what a standard deviation represents...
One standard deviation away from the mean in
either direction on the horizontal axis (2)
accounts for somewhere around 68 percent of
the samples. Two standard deviations away
from the mean (4), account for roughly 95
percent of the samples. Three standard
deviations (6) account for about 99 percent of
the samples. If this curve were flatter and more
spread out, the standard deviation would have to
be larger in order to account for the 68 percent
so that's why the standard deviation can tell you
how spread out the examples in a set are from
the mean.

This is useful in manufacturing as it tells us a lot
about the quality of what you are making and how the equipment used in the process is behaving. So if
for example you were monitoring the values of electrical resistors or the diameter of pistons being
produced by machinery, a small standard deviation will tell us that they are being produced very
accurately and close to the mean. Suppose all the samples between 4 and 6 are rejects. The larger the
value of , the more rejects. Also if the mean of the sample is moving as time goes on, it means that more
samples will be rejected on one side than the other and indicates that the something in the machine (e.g.
the grinding wheel) is wearing away.
D.J.DUNN 10
11. CALCULATION OF STANDARD DEVIATION for UNGROUPED DATA

Ungrouped data is presented in a table listing the value of each sample. If the number of samples is large,
this becomes a large table but it is probably best to use this method with small numbers of samples.

Standard deviation = Square root of the mean of the variables squared.
The variance is denoted S =
2
( )
1 n
x x
S
2
2

= =

n = number of samples
You might find it better to arrange your tables in columns rather than rows. Lets look at another
example.


WORKED EXAMPLE No.2

The following is a table of lead concentration in the blood of a group of people. Calculate the mean
and the standard deviation.

Sample Resistance (Ohms) Difference from mean Differences squared
1 119 -1.8 3.24
2 120 -0.8 0.64
3 120 -0.8 0.64
4 121 +0.2 0.04
5 122 +1.2 1.44
6 119 -1.8 3.24
7 119 -1.8 3.24
8 122 +1.2 1.44
9 123 +2.2 4.84
10 123 +2.2 4.84
Totals 10 1208 23.6

Mean = 1208/10 = 120.8

The sum of the squares of the differences (or deviations) from the mean, 23.6, is now divided by the
total number of observation minus one, to give the variance.

VARIANCE

( )
2.622
9
23.6
1 n
x x
S
2
2
= =

= =



Finally, the square root of the variance provides the standard deviation:

= 2.622

= 1.619 Ohms

D.J.DUNN 11


SELF ASSESSMENT EXERCISE No.2

1. The hardness of ten steel samples was measured and the results were as follows.

Sample 1 2 3 4 5 6 7 8 9 10
Hardness 90 92 95 91 98 102 97 92 95 99

Calculate the mean and the standard deviation. Answer 95.1 and 3.9

2. The thickness of 20 steel strips was measured in mm and tabulated as shown.

Sample 1 2 3 4 5 6 7 8 9 10
Thickness 19.8 19.9 19.9 20.1 20.1 19.9 20.2 19.7 19.7 19.9


Calculate the mean and the standard deviation. Answer 19.92 and 0.168




D.J.DUNN 12

12. CALCULATION OF STANDARD DEVIATION for GROUPED DATA

Grouped data is presented in tables showing the bands and the frequency and is more likely to be used
with large numbers of samples.

Consider a normal distribution curve. The mean occurs at the middle. The deviation from the mean at any
point is d. Next consider the graph of d plotted against f and further the graph of d
2
plotted against f. On
this last graph we find the mean d
2
as follows.


The area of the graph may be found from the mid-ordinates. ) .....d d d w(d A
2
n
2
3
2
2
2
1
+ + + =
The mean height of the graph is the variance
n
2
n
2
3
2
2
2
1
f
) .....d d d w(d
S
+ + +
=
w is the width of each strip = f
n
/n and n is the number of strips. The mean height is then
n
) .....d d d (d
nf
) .....d d d (d f
S
2
n
2
3
2
2
2
1
n
2
n
2
3
2
2
2
1 n
+ + +
=
+ + +
=
In general
n
d
S

=
2
For reasons not explained here, n-1 is often used instead of n on the bottom line. d
is the deviation d = x - x
( )
1 n
x x f
S
2
2

= =

is the standard deviation.
It can be shown that the formula simplifies to
2
2
2
f
fx
f
fx

D.J.DUNN 13

WORKED EXAMPLE No.3

The following is a grouped set of data for visits made to the doctor by a sample of children.

Visit to Doctor No.of Children Total Visits Cumulative x
2
fx
2

x f fx
0 2 0 2 0 0
1 8 8 10 1 8
2 27 54 64 4 108
3 45 135 199 9 405
4 38 152 351 16 608
5 15 75 426 25 375
6 4 24 450 36 144
7 1 7 457 49 49

Totals f = n = 140 f x = 455 fx
2
=1697

Mean number of visits = 455/140 = 3.25.


1.25
1.55
140
455
140
1697
f
fx
f
fx

2
2
2
2
=
=

Some texts give the formula as
1.25
1.57
140
455
139
1697
f
fx
1 f
fx

2
2
2
2
=
=

This does not make much difference so long as the total number of samples is very small.










D.J.DUNN 14

WORKED EXAMPLE No.4

The hardness of 143 samples of steel is measured and grouped into bands as shown. Calculate the mean and
standard deviation. The figures of 17.5 and 21.5 result from one sample being exactly 91 units and so half is
allocated to each band.


Range mid point freq. fx acc f x
2
f x
2

x f
89-91 90 17.5 1575 17.5 8100 141750
91-93 92 21.5 1978 39 8464 181976
93-95 94 32 3008 71 8836 282752
95-97 96 38 3648 109 9216 350208
97-99 98 17 1666 126 9604 163268
99-101 100 17 1700 143 10000 170000
Totals 143 13575 1289954

Mean = 13575/143 = 94.93



2.99
8.939
143
13575
143
1289954
f
fx
f
fx

2
2
2
2
=
=

It is of interest to note that in this population, we get a very different answer using the other formula.


8.51
72.46
143
13575
142
1289954
f
fx
1 f
fx

2
2
2
2
=
=

D.J.DUNN 15


SELF ASSESSMENT EXERCISE No.3

1. The accuracy of 100 instruments was measured as a percentage and the results were grouped into bands of 1%
as shown. Calculate the mean and the standard deviation.

Range Mid freq
61.5-62.5 62 1
62.5- 63 2
63.5- 64 3
64.5- 65 4
65.5- 66 8
66.5- 67 12
67.5- 68 13
68.5- 69 18
69.5- 70 14
70.5- 71 10
71.5- 72 5
72.5- 73 4
73.5- 74 3
74.5- 75 2
75.5-76.5 76 1


Answers 68.88 and 2.74%

2. The breaking strengths of 150 spot welds was measured in Newton and grouped into bands of 20 N as shown.

Range f
160-10 2
180-200 6
200-220 10
220-240 28
240-260 50
260-280 31
280-300 15
300-320 8

Calculate the mean and the standard deviation. (Answers 251.47 N and 29.04 N)