Vous êtes sur la page 1sur 12

Introduction

When you think


Wh hi k off simulation,
i l i
what comes to mind?

CHAPTER 9: Simulation

Introduction Introduction
Computer simulation:

A simulator is an experimental device that A quick & inexpensive way to acquire


“acts like” or simulates the system of knowledge concerning a problem that is
interest. usually gained through experience

Computer models can be a type of C


Creates
t an environment
i t to
t obtain
bt i
simulator when they are developed to information in a quick, cost-
cost-effective
evaluate a business problem. and low risk manner about alternative
actions

1
Simulation should be used when: Excel Features & Functions
FEATURES
• Data Analysis: Descriptive Statistics
1. Analytical models are difficult or • Data Analysis: Histogram

impossible to develop because of FUNCTIONS


complicating factors. • RAND: generates a random real number
• INT: rounds a number down to the nearest integer
• NPV: calculates net present value of an investment
2. We want to predict something other than • SQRT: returns a positive square root
• AVERAGE: returns the arithmetic mean
average or “steady
steady-state”
state” (long
(long-run)
run) • VLOOKUP: searches for a value in the first column of a table array and
returns a value in the same row from another column in the table array
behavior. • COUNTIF: counts the number of cells within a range that meet defined
criteria

Terminology Monte Carlo Simulation

1. Construct a Cumulative Probability


• Simulation Models
Di t ib ti
Distribution
• Probability Distributions
2. Create Random # Distribution from
• Cumulative Probability Distributions
Cumulative Probability Distribution
• Random Variables/Random Numbers(RN)
−discrete
di vs. continuous
i 3 Generate Random Numbers
3. (RAND)
• Expected Value 4. Project a Random Observation of the
Variable (VLOOKUP)

2
Probability Distributions Continuous Distributions

• Normal distribution
Discrete: integer values (whole numbers) − Most widely used
− Conditions:
Continuous: fractional or decimal values ƒ Some value is most likely (mean)
ƒ Symmetrical around mean
(infinite number of values) ƒ More likely to be near mean
Demand % Cumu %
Cumulative • Uniform distribution
Distributions 8 .10 0
− All values between min & max occur
9 .20 .10 with equal likelihood
10 .30 .30 − Conditions:
ƒ Min value is fixed
11 .20 .60
ƒ Max value is fixed
12 .10 .80 ƒ All values occur with equal likelihood
13 .10 .90

To generating random variables:

Draw random sample from given


probability distribution.
distribution

• Random variable: a variable whose


outcome is not known with certaintyy
Generate random numbers

3
To generating random variables: Monte Carlo Simulation
13 (10.0%) 14 (10.0%)
Draw random sample from given
probability distribution.
distribution 15 (10.0%)
12 (20.0%)
• Probability distribution: describes the
behavior of a random variable by defining
its:

ƒ Central Tendency: Mean (expected value)


ƒ Variability: Standard Deviation 10 (30.0%)
11 (20.0%)
ƒ Limit: Upper & lower (min/max)
ƒ Nature: Discrete vs Continuous

Monte Carlo Simulation Probability


13 (10.0%)
Distributions
14 (10.0%)

15 (10.0%)
12 (20.0%) The Toss of a Fair Coin

TAILS HEADS
10 (30.0%) 50% 50%
11 (20.0%)

4
Probability
Distributions

The Roll of a Fair Die

SIX ONE
17% 16%
FIVE TWO
1 %
17% 16%

FOUR THREE
18% 17% Generating random numbers

Generating Random Numbers


Continuous random variable
requirements:
1. Ability to generate decimal numbers

2. Understand cumulative
distribution of the random variable
We’ll use Excel’s Rand function to be generated
to generate random numbers

5
MGS3100 Student Age Distribution
Generating Random Numbers Fall 2009
Continuous Distribution function

Demand % Cumu % Random # Limit:


8 .10 .10 0.0 – 0.09999 Minimum Age = 19; Maximum Age = 40
9 .20 .30 0.1 – 0.29999
10 .30 .60 0.3 – 0.59999
Prob (D<=x) Central Tendency:
11 .20 .80 0.6 – 0.79999
M
Mean = 24 Median
M di = 22 M d = 21
Mode
12 .10 .90 0.8 – 0.89999
13 .10 1.00 0.9 – 0.99999
Nature:
Discrete vs Continuous?

MGS3100 Student Age Distribution


Fall 2009 Classroom Exercise
Age # % Cumulative % Question 9
9--4: Sally’s Solar Car
19 1 2.5% 2.5%
20 7 17.5% 20.0% Sales/
21 10 25.0% 45.0% Week # Weeks What probability
22 7 17.5% 62.5% distribution & cumulative
0 2 probability distribution
23 1 2.5% 65.0%
1 5 would you assign to
24 2 5.0% 70.0% demand so that the
25 1 2.5% 72.5% 2 12
probability of a particular
26 2 5.0% 77.5% 3 16 demand in the simulation
27 1 2.5% 80.0% 4 8 is equal to the relative
frequency of that demand
>28 8 20.0% 100.0% 5 7 over the last 50 weeks?

6
Simulate Demand with Random
Generating Random Numbers Numbers
Continuous random variable 3. Assign random numbers, using an interval from 0
requirements: to 1,
1 to formulate a continuous distribution

Demand % Cumu % Random#


1. Ability to generate decimal numbers 8 .10 .10 0.0 – 0.09999
9 .20 .30 0.1 – 0.29999
2. Understand cumulative 10 .30 .60 0.3 – 0.59999
distribution of the random variable 11 .20 .80 0.6 – 0.79999
12 .10 .90 0.8 – 0.89999
to be generated
13 .10 1.00 0.9 – 0.99999
The probability of drawing a number between .9
and .99999 is 1 out of 10 or 0.1 (10%).

Simulate Demand with Random


Numbers Expected Values
Multiply profit (in this example) by respective
4. Generate random numbers and project probabilities to get expected profit
d
demandd
Random
Number Demand
Demand % Cumu % Random# 0.7612 11
0.7184 11
8 .10 .10 0.0 – 0.09999
0.1437 9
9 .20 .30 0.1 – 0.29999 0.9471 13
10 .30
30 .60
60 0.3 0 59999 0.1574
0 3 – 0.59999 0 1574 9
0.3308 10
11 .20 .80 0.6 – 0.79999 0.2962 9
0.5629 10
12 .10 .90 0.8 – 0.89999
0.1832 9
13 .10 1.00 0.9 – 0.99999 0.7194 11
0.2679 9
0.7444 11 Table 2, p187

7
All-Ways-Open Market All-Ways-Open Market
Q13, p216
1. Construct a Cumulative Probability
Di t ib ti
Distribution

Sales Week Prob. Cumu Random #s


40 4 0.08 0.08 0.00-0.7999
41 10 0.20 0.28 0.80-0.2799
42 12 0.24 0.52 0.28-0.5199
43 9 0.18 0.70 0.52-0.6999
44 8 0.16 0.86 0.70-0.8599
45 7 0.14 1.00 0.86-0.9999
Total 50

Review:
Monte Carlo Simulation Recap Generating Random Numbers

1. Construct a Cumulative Probability


Continuous random variable
Di t ib ti
Distribution requirements:

2. Create Random # Distribution from 1. Ability to generate decimal numbers


Cumulative Probability Distribution
2. Understand cumulative
3 Generate Random Numbers
3.
distribution of the random variable
4. Project a Random Observation of the to be generated
Variable (VLOOKUP)

8
Monte Carlo Simulation Recap Generating Random Numbers

1. Construct a Cumulative Probability


Discrete random variable
Di t ib ti
Distribution requirements:

2. Create Random # Distribution from 1. Ability to generate whole numbers


Cumulative Probability Distribution
2. Understand distribution of the
3 Generate Random Numbers
3.
discrete random variable to be
4. Project a Random Observation of the generated
Variable

Generating Random Numbers Generating Random Numbers


Discrete Distribution Function Discrete Distribution Function
Excel 2007: The RANDBETWEEN function Excel 2003: The INT function used in
can be used to generate discrete distribution conjunction with the RAND function
generates discrete distribution

=RANDBETWEEN(8,12) returns a discrete Values for RAND() =INT(8+5*RAND())


uniform distribution of integers between =INT(8 + 5 * RAND()) 0 <=RAND() <0
<0.2
2 8
8 & 12 returns a discrete 0.2 <RAND() <0.4 9
uniform distribution 0.4 <RAND() <0.6 10
of integers between 8 0.6 <RAND() <0.8 11
& 12 0.8 <RAND() <1.0 12

9
Startup Costs $150,000
Sales Price $ 35,000
Fixed Costs (per year) $ 15,000
Variable Costs (per year) 75% of revenues
Airbus: Capital
Budgeting for Tax depreciation: $10,000/yr over 4 yr product life

New Product Salvage value: Estimated to be $0.00


Line
Cost of capital: 10%

Tax rate: 34%

Demand: ??, but to get started we’ll assume 10


planes/year for 4 yr product life
P. 167

AIRBUS: NPV Model w/Random


AIRBUS: NPV Model Demand

Assumptions:
• Yearly demand will be either 8, 9, 10, 11, or 12
The model • Each demand value is equally likely to occur
uses Excel’s
NPV function
to calculate We will include RANDBETWEEN function to sample
Net Present from a discrete uniform distribution on the integers
Value 8, 9, 10, 11, 12

We will generate multiple simulation trials

10
AIRBUS: NPV Model w/Random
Demand AIRBUS: Evaluate the Proposal

We modify We need to answer two questions about the NPV


the NPV distribution:
model to 1. What is the mean or expected value of the NPV?
generate
random 2. What is the probability that the NPV assumes a
demands for negative value?
each year We’ll create a simulation model to answer these
questions.

Use Data Table to run the simulation automatically


and capture the resulting NPV in a separate table.

AIRBUS: Evaluate the Proposal AIRBUS: Evaluate the Proposal


1. What is the mean or expected value of the NPV? 2. What is the probability that NPV assumes a
g
negative value?

• What is the best possible outcome?

We review the Distribution of Outcomes:


• What is the worse possible outcome? The Frequency Table

11
AIRBUS: Evaluate the Proposal AIRBUS: Evaluate the Proposal
3. How confident are we in these results? 4. Would we be more confident if we ran more trials?

• Increasing
I i the
th # off trials
t i l will
ill give
i better
b tt NPV
Confidence intervals use the following statistics: estimate, but there will always be differences
1.00 =68% 1.96 = 95.0% 3.00 = 99.7% between the average & true expected return

To compute a 95% 1.96 ∗ std


confidence interval: mean ± • Simulations can provide useful information
n on the distribution results

• Simulation results are sensitive to


assumptions affecting input parameters.

12

Vous aimerez peut-être aussi