Probability and Statistics for

Engineers
Course 461601
Applied Statistics
Course 400516

1
 Course Description
This course considers the fundamentals of statistical and probabilistic
analysis. Theory of probability and random variables, and
Probability/statistical distributions are discussed. Estimation of
parameters and testing hypothesis will be used in the decision making
process. Data fitting techniques will be used in regression analysis
Text Books
• Applied Statistics and Probability for Engineers, Douglas Montgomery
and George Runger
• Probability, Statistics, and Reliability for Engineers and Scientists, By
Ayyub and McCuen

2
Topics in the course
Week Subject
Lecture Subject
1-2 Introduction. Data description and treatment, Fundamentals of probability
3-5 Probability distributions for Discrete Random Variables
5-7 Probability distributions for Continuous Random Variables
8-9 Fundamentals of statistical analysis and estimation theory
10-12 Test of hypotheses and Analysis of Variance
13-14 Regression analyses and data fitting techniques (Linear and nonlinear)

15 Introduction to Reliability Analysis

16 Final exam

3
Evaluation Policy
• This is a remote e-learning course.
• Interaction and Participation will be through ZOOM environment
• Communication will be through email: m.m.dwaikat@najah.edu and through
MOODLE of An-Najah Nat. Univ.
• Phone-calls are discouraged except for ultimate necessity (e.g. no internet access
for exam/homework submission)
• The scores of the course will be assessed using:
• Restricted-Time Exam 20%
• Series of Home-works / mini-Projects 40%
• Final exam 40%

4
 Role of Statistics in Engineering
• The field of statistics deals with the collection, presentation, analysis, and
use of data to make decisions, solve problems, and design products and
processes.

The Engineering Method for

solving any physical problem

5
 Data vs Information vs Knowledge
• Data are raw unorganized numbers/shapes/figures concerning some measurable or
quantifiable phenomenon.
• Pieces of data represent real physical facts!
• When the data are organized and processed into a presentable “useful” form, they
are called Information.
• The word “useful” is relative, and it generally means “ready to be used for a
decision process or validating a piece of “knowledge”
• Knowledge is the whole or partial understanding of a phenomenon or behavior.
• Knowledge generate connects the pieces of information through cause-and-effect
or relational/reasoning statements
• Example: Height (H) and Life Span (L) of a group of people is Data. A graph showing
the correlation coefficients between H and L for many groups are pieces of
information. Finding a reasoning for the trend between H and L is a piece of
6
knowledge.
 Data Collection
Three basic methods of collecting data are
• A retrospective study using historical data
e.g. seismic peak accelerations. Failure types in an earthquake
Events had occurred. Many variabilities. Biased selection, or lack of data
• An observational study
e.g. Percentage of columns with cross-sectional area < Designed Area.
Events are already in progress or they happen without control
• A designed experiment/Simulation
Controlled environment. Used to study effect of some parameter variation.
e.g. Effect of using spiral stirrups on ductility!
7
Role of Bias in the Data Collection
• Sometimes, collected data reflect only part of the
truth.
• In the 1940s, the Allies tried to improve their
warplanes and make them more reliable in battle by
fortifying them against gun fire.
• So, they started collecting data about the warplanes
that come back from battles.
• They marked the most hit regions in the planes, and Example of hit regions in a
warplane coming from battle
most Engineers suggested to put extra material on
those regions.
• Few engineers suggested quite the opposite!! …
Mechanistic vs. Empirical
Deterministic vs. Stochastic
• Some times a phenomenon can be described by an exact physical law. This is called mechanistic
model,
e.g. Elastic buckling force of a compression member: Pn = π2*E*Ag/ (kL)2
e.g. Elastic stress due to bending in thin beam: s = M y / I
• If the values of E, Ag, L , M, ... etc, are characterized by a single value (e.g. their mean value), then
this model is deterministic
• if they are probabilistic, or random, then this model is stochastic or probabilistic.
• Some times, it is difficult or impossible to derive a clear physical law to describe an event,
e.g. the time for a concrete beam to fail under fire. Tn = Function of (L, B,H, F’c, f, …, etc)
In this case, a regression (or fitted model) for the phenomenon is used based on many experiments
(observations)
i.e.: Tn = a1*L+a2*B+a3*H+a4*F’c + … , where a1, a2, a3..etc are “fitting” or “regression” coefficients
determined from fitting of many “observed” failing times.
Such model is called an “empirical” model 9
 Sample and Population
• A population represents all the possible values for an event, whether such
events had all occurred or partially occurred or yet to occur, whether
existed, existing, or yet to exist.
• A sample represents a subset of possible events within a population.
• Any information derived based on a sample is called a “statistic”
• Any information derived based on a population is called a “parameter”
• Generally, a parameter of a population can not be known exactly. It can
only be “estimated” or “inferred” using inferential or estimation theory.

10
 Types of Statistical Analysis
• Descriptive Statistics
Deals with summarizing information about data from a sample or samples,
regardless of their origin, and with no regard to their relation to population
e.g. Central measures, and dispersion measures…etc.
• Inferential [Inductive] Statistics
Deals with using information from samples to draw conclusions about population.
e.g. point estimation, interval estimation, testing of hypotheses…etc.
• Generally, inferential statistics is based on Probabilistic analysis.

11
Summarizing Data / Central Measures
• These measure the central tendency of the data
• The Arithmetic Sample Mean

• The median = The value above which is half the number of all
values in a set.
• The mode = The most frequent value within a sample.

12
Summarizing Data / Dispersion Measures
•  These values measure how much the values tend to scatter.
• The sample variance

• The sample standard deviation

• The coefficient of variation (C.o.V), also known as relative standard

deviation (RSD):
For population CoV = sigma/mu
13
Summarizing Data / Dispersion Measures
• Percentiles: Represent the value which divided the number of events in a
sample in a certain ratio above and below it.
• E.g. 25-percentile  25% of the values are below the value, also called First
Quartile [Q1]
• 75-percentile  a value which has 75% of all events below it. Also called Third
Quartile [ Q3]
• Note the median is the second quartile [Q3]
• Interquartile range [IQR] = Q3 – Q1
• The IQR measures the range of middle 50% of the values.
• E.g.: Take the sample space S = {67, 69, 70, 71, 74, 77, 78, 82, 89} , which
represents the crushing capacity of a material in N.
• The median = Q2 = the 50th Percentile 74
• Q1 or the 25th percentile = (69+70)/2 = 69.5, while Q3 = (78+82)/2 = 80
• IQR = 80-69.5 = 11.5 14
 Probability Theory
Three Approaches to define the probability of an event
• CLASSICAL APPROACH: If an event “A” can occur in h different ways out of a
total number of n possible ways, all of which are equally likely, then the
probability of the event A is P(A) = h/n
We have a problem with defining “equally likely”
• FREQUENCY APPROACH. If after n repetitions of an experiment, where n is very
large, an event “A” is observed to occur in h of these, then the probability of the
event is P(A) = h/n. This is also called the empirical probability of the event.
We have a problem with defining “large n”
• Since both approaches have drawbacks, there is an AXIOMATIC APPROACH of
defining probability.

15
Axioms of Probability
• Axiomatic Definition: Suppose we have a sample space S. To
each event A in S, we associate a real number P(A). Then P is
called a probability function, and P(A) the probability of the
event A, if the following axioms are satisfied:
• Axiom 1: For every event A in S: 1 ≥ P(A) ≥ 0
• Axiom 2: For the sure or certain event S : P(S) = 1
• Axiom 3: For any number of mutually exclusive events A1 , A2, in
S, P(A1ꓴ A2) = P(A1) + P(A2) and P(A1ꓵ A2) = 0,
where A1 , A2 are mutually exclusive if A1ꓵ A2= φ
16
Resulting Theorems in Probability

17
Axiomatic Approach

18
Conditional Probability and Independence

19
Bayes’ Theorem

20
 Example
• Three Companies [A, B, and C] produce table lamps. The companies produce 30%, 20%
and 50% of the lamps for A, B, and C, respectively. If the percent of defect lamps in A is
10%, but in B it is 30% and in C it is 7%. If we bought a defect lamp, what is the
probability that this lamp is from company C.
Summary Table
• Here we know the probabilities of Defects per Company Prob of a lamp from Defect lamp ratios per
company company
company [P(D|Comp)]. What is asked for is the
A 0.3 = P(A) 0.1 = P(D|A)
probability the lamp is from company C if it is B 0.2 = P(B) 0.3 = P(D|B)
given that it is a defect, i.e. P(C|D) C 0.5 = P(C ) 0.07 = P(D|C)

 P(C|D) =

 
0.1*0.3+0.3*0.2+0.07*0.5=0.125
P(C|D) = 0.035/0.125 = 0.28 = 28%
21
 Permutations and Combinations
• If we have “n” objects and we want to choose “r” at once and at random
from n, then we can do this in nPr possible ways, where:

e.g. Let S = { A,B,C,D}, and we want to choose 2 letters at once, then we have the
possibilities: {A,B},{B,A}, {A,C},{C,A}, {A,D},{D,A}, {B,C},{C,B} {B,D},{D,B}, {C,D},{D,C} = 12
possibilities. Here n = 4, and r = 2 , thus. 4P2 = 4*3 = 12, Or 4P2 = (4*3*2*1)/(2*1)=12.
• If we do not care about the “order” of selecting the “r” cases, then we use
the combination rule nCr

e.g., in the previous example, we want to choose any two letters with no importance for
22