lOMoARcPSD|43106416
Chapter 16 notes
Business Analytics (University of Pretoria)
Scan to open on Studocu
Studocu is not sponsored or endorsed by any college or university
Downloaded by Thando Thando (tthando273@[Link])
lOMoARcPSD|43106416
Chapter 16: Risk Analysis and Simula琀椀on
Overview of Prescrip琀椀ve Analy琀椀cs:
Prescrip琀椀ve analy琀椀cs: a process of using analysis tools such as simula琀椀on and op琀椀miza琀椀on
to improve decision-making. It can help a decision-maker determine the best course of
ac琀椀on among di昀昀erent alterna琀椀ves. (Be able to de昀椀ne and explain, not do)
➢ Some people regard prescrip琀椀ve analy琀椀cs as a very broad term that encompasses all
analy琀椀cal techniques with an overall goal of improving business decision-making.
➢ Prescrip琀椀ve analy琀椀cs uses simula琀椀on and op琀椀miza琀椀on techniques to quan琀椀fy the
e昀昀ect of di昀昀erent possible ac琀椀ons by a decision-maker to help make a more
informed decision.
The 3 types of analy琀椀cs techniques that are used to extract value from data and make be琀琀er
business decisions:
Descrip琀椀ve analy琀椀cs: refers to gathering, organizing, tabula琀椀ng, and visualizing data to summarize
“what has happened”.
Predic琀椀ve analy琀椀cs: refers to using historical data to predict “what could happen in the future”.
Examples: past sales, revenue stream, produc琀椀on
Prescrip琀椀ve analy琀椀cs: refers to using simula琀椀on and op琀椀miza琀椀on algorithms to provide advice on
“what we should do”.
➢ Simula琀椀on: an a琀琀empt to imitate a real-world process that examines several possible
scenarios.
➢ Op琀椀miza琀椀on: an a琀琀empt to 昀椀nd an op琀椀mal way to achieve an objec琀椀ve under given
constraints, such as limited capaci琀椀es, 昀椀nancial resources, and compe琀椀ng priori琀椀es.
Risk and Risk Analysis (Assessing risk):
Risk: the possibility or probability that an undesirable outcome may occur.
Risk analysis: a process of examining and assessing di昀昀erent scenarios to be琀琀er understand
the possibility of both intended and undesired outcomes. Also called scenario analysis.
Conduc琀椀ng Risk Analysis:
We typically perform risk analysis by considering 3 alterna琀椀ves: most likely, pessimis琀椀c
(worst case), and op琀椀mis琀椀c (best case) scenarios, allowing us to evaluate a range of
outcomes that occur most frequently. Many prac琀椀琀椀oners use historical data to help es琀椀mate
the worst and best outcomes. We can assess and model risk using random variables and risk
analysis. We use random variables to capture the inherently uncertain nature of processes
to make be琀琀er-informed decisions.
Downloaded by Thando Thando (tthando273@[Link])
lOMoARcPSD|43106416
Using Random Variables to Model Risk and Uncertainty: (EXAM)
Random variable: a func琀椀on that assigns numerical values to the outcomes of an
experiment.
➢ Discrete random variable: assumes a countable number of dis琀椀nct values. For
example, the number of students receiving the grade of A.
➢ Con琀椀nuous random variable: characterized by uncountable values in an internal. For
example, the return on an investment or the height of individuals.
Using Excel to generate Random Observa琀椀ons from a Discrete Probability Distribu琀椀on:
of the most relevant discrete probability distribu琀椀ons in most business and nonbusiness
applica琀椀ons are: Binomial and Poisson distribu琀椀ons.
Binomial random variable Poisson distribu琀椀on
The number of successes achieved in the n trials Is useful when we are interested in 昀椀nding the
of the Bernoulli process. number of occurrences (successes) of a certain
event over a given interval of 琀椀me or space.
Example: the number of customers, in a sample
of 100, whose loan applica琀椀on will be approved Example: the number of customers arriving at a
by a bank. retail store between 10 a.m. and noon on a
weekday.
Another important discrete probability distribu琀椀on is the Discrete Uniform Distribu琀椀on:
A discrete uniform random variable has a 昀椀nite number of speci昀椀ed values, with each value being
equally likely.
When rolling a fair six-sided die, the number rolled is an example of a discrete uniform random
variable.
Using Excel to generate random observa琀椀ons from a Con琀椀nuous probability distribu琀椀on:
Normal distribu琀椀on is the most extensively used probability distribu琀椀on for sta琀椀s琀椀cal work,
analy琀椀cal modeling and accurately describes numerous random variables of interest. Can be
posi琀椀ve and nega琀椀ve. Examples include the amount of rainfall during a plan琀椀ng season or
students’ exam scores. Two other important con琀椀nuous distribu琀椀ons include:
Con琀椀nuous uniform distribu琀椀ons:
Con琀椀nuous uniform distribu琀椀on is also referred to as rectangular distribu琀椀on as it represents a
constant probability within a speci昀椀ed range.
Examples: delivery 琀椀me of an appliance, the arrival 琀椀me of an elevator, or the scheduled 昀氀ight 琀椀me
between des琀椀na琀椀ons.
Exponen琀椀al distribu琀椀on:
The exponen琀椀al distribu琀椀on is appropriate when the data are believed to be generated from a
nonsymmetric distribu琀椀on. it has a “memoryless” feature that implies a constant failure rate
regardless of the history.
It is also related to the Poisson distribu琀椀on.
Examples: 琀椀me between text messages or between customer purchases. It is also used in modeling
failure 琀椀mes such as the life of an appliance.
Downloaded by Thando Thando (tthando273@[Link])
lOMoARcPSD|43106416
Using Excel to generate random observa琀椀ons from a Triangular distribu琀椀on: (NB EXAM)
Triangular distribu琀椀on: a simplis琀椀c representa琀椀on of a con琀椀nuous distribu琀椀on with a
triangular shape and is de昀椀ned by 3 parameters: the minimum value, the maximum value,
and the mode(Occurs the most).
➢ Some琀椀mes the minimum value represents the worst-case scenario (Ex. Lowest pro昀椀t
level) and the maximum value represents the best-case scenario (Ex. The highest
pro昀椀t level).
➢ In some situa琀椀ons, the reverse could occur, such as where manufacturing cost where
the minimum amount represents the best case.
➢ Some琀椀mes we consider the average value to represent the most likely scenario.
➢ When the triangular distribu琀椀on has a symmetrical shape, the mode and mean
values are the same.
Monte Carlo Simula琀椀on: (EXAM), DEF, PURPOSE ANS STEPS
Monte Carlo simula琀椀on: used to model probabilis琀椀c (stochas琀椀c) processes based on
random variables. By recrea琀椀ng a real-world process, Monte Carlo simula琀椀on is used to
understand the impact of risk and uncertainty in a wide variety of applica琀椀ons.
This technique relies on random variables to mimic the odds of all possible outcomes,
Monte Carlo simula琀椀on is also called probabilis琀椀c or stochas琀椀c simula琀椀on, as opposed to a
determinis琀椀c process that can be accurately described by a formula and outcome can be
precisely es琀椀mated.
Downloaded by Thando Thando (tthando273@[Link])
lOMoARcPSD|43106416
Steps to perform a Monte Carlo simula琀椀on:
Not required to do the Monte Carlo simula琀椀on
1. Data collection
• We usually gather histrorical data to make an assumption about the underlying
probability distribution of a random varaible.
• Histrorical data can also be used to estimate the empirical probability distribution.
2. Random number generation and assignment
• To model risk and uncertainty in a Monte Carlo simulation, we rely on computer
software to generate random numbers that correspond to the appropriate probability
distribution.
3. Model formulation
• The next crucial step is to construct a quantitative model that represents the
relationship among relevant variables.
• The model formulation varies among different applications.
• regardless of the context or underlying situations, the Monte Cralo formulation needs
to capture the uncertainty with random variables, in relation to the outcome.
4. Performing and repeating of the analysis
• Based on the quantitative formulation, we can perform and repeat the analysis aand
examine the results.
• If the simulation process is repeated many times, relevant information such as
average values, most likely, best and worst cases, and a range of outcomes can be
gathered to inform decision makers about the appropraite course of actions.
5. Interpretation of the results and deployment
• After performing the simulation, we consider the results and develop actionable
recommendations.
• we aslo need a strategy for implementing the recommendations and for monitoring
and documenting the actual outcomes.
Downloaded by Thando Thando (tthando273@[Link])