Académique Documents
Professionnel Documents
Culture Documents
©
HOCK international - 2004 1 ©
HOCK international - 2004 2
Part 1 CMA – Section D: Quantitative Methods Part 1 CMA – Section D: Quantitative Methods
©
HOCK international - 2004 3 ©
HOCK international - 2004 4
The four components of a time series that affect its historical The two basic forecasting methods are:
values are:
1) Time Series Analysis, and
1) Trend – A gradual shifting to a higher or lower level.
Example: long-term sales growth. 2) Causal Forecasting methods.
A moving average is the average of the most recent data in the Time series methods are used in forecasting in three ways:
time series. A four-week moving average of sales is the average of the 1) Smoothing, including moving averages, weighted moving averages,
sales each week for the most recent four weeks. Each time a new and exponential smoothing,
value becomes available, it replaces the oldest value.
2) Trend projection using the high-low points method or simple linear
regression, and
A weighted moving average is a moving average that uses different
3) Trend projection adjusted for seasonal influence using
weights for each value. For example, more recent historical values
deseasonalized data in a simple linear regression.
might be given more weight than those given to older values.
Part 1 CMA – Section D: Quantitative Methods Part 1 CMA – Section D: Quantitative Methods
©
HOCK international - 2004 5 ©
HOCK international - 2004 6
Part 1 CMA – Section D: Quantitative Methods Part 1 CMA – Section D: Quantitative Methods
©
HOCK international - 2004 7 ©
HOCK international - 2004 8
The advantages of exponential smoothing are:
Exponential smoothing is a special type of weighted moving average.
• It is inexpensive because it does not require a lot of historical data. We forecast the value for the next period using the most recent period’s
It requires only the current period’s actual and the current period’s actual value and the most recent period’s forecasted value that has
forecast, so data storage requirements are minimized. been forecasted using exponential smoothing.
• It is a simple concept. The most recent period’s actual value and the most recent period’s
• It is quite powerful because of its weighting process. forecasted value are each weighted (with their total weights being
equal to 1). This weighted average becomes the next period’s forecast.
The disadvantages of exponential smoothing are:
• The forecast will lag if the trend increases or decreases over time.
Exponential smoothing as a forecasting technique is most useful
• It does not account for dynamic changes that occur.
when the time series is stable, without many fluctuations.
• Its forecasts require constant updating to respond to new
information.
• Its usefulness is limited, because it is most useful when the time
series is stable, without many fluctuations.
The two assumptions in using either the high-low points method or Trend projection, using the high-low points method or simple regression
simple linear regression analysis to make a trend projection are: analysis, is most appropriate when a time series is not stable but
instead is increasing or decreasing consistently.
1) Variations in the dependent variable (what we are forecasting) are
explained by variations in one single independent variable (time, in
a time series).
©
HOCK international - 2004 9 ©
HOCK international - 2004 10
Part 1 CMA – Section D: Quantitative Methods Part 1 CMA – Section D: Quantitative Methods
©
HOCK international - 2004 11 ©
HOCK international - 2004 12
The coefficient of correlation, r, measures the relationship between
two or more variables. It shows how closely connected the variables are
The coefficient of determination, r 2, is the square of the coefficient
and the extent to which a change in one variable will result in a change in
of correlation. It represents the percentage of the total amount of the other variable.
change in the dependent variable that can be explained by changes in
The coefficient of correlation is a numerical measure that measures both
the independent variable. In other words, it is the amount of change in
the direction (positive or negative) and the strength of the linear
the y value of the regression line. association between the value of x and the value of y. It is expressed as
R2 is expressed as a number between 0 and 1. If the coefficient of a number between -1 and +1.
determination is high, the data points will all lie close to the trend line, A correlation coefficient of +1 means there is a perfect positive (upsloping)
and vice versa. linear relationship between each value for x and its corresponding value
for y. A correlation coefficient of -1 means there is a perfect negative
(downsloping) relationship between each value for x and its corresponding
value for y. A correlation coefficient close to zero may mean there is very
little relationship between the variables, or it may mean there is a strong
relationship but it is not a linear relationship.
The coefficient of correlation is used to determine whether trend projection
would be meaningful.
Causal forecasting uses some other value to determine a value we The equation of a linear regression line is:
are forecasting. y^ = ax + b
Causal forecasting is used when there is an identified cause and effect Where:
relationship between the value we are forecasting and some other y^ = the value of y on the regression line corresponding to
value. each value of x
a = the slope of the line
b = the y intercept, or the value of y when x is 0
x = the value of x on the x axis that corresponds to the value
of y on the trend line
©
HOCK international - 2004 13 ©
HOCK international - 2004 14
Part 1 CMA – Section D: Quantitative Methods Part 1 CMA – Section D: Quantitative Methods
©
HOCK international - 2004 15 ©
HOCK international - 2004 16
Benefits and advantages of regression analysis:
• Regression analysis is a quantitative method and thus it is objective.
Real Value = Nominal Value A given data set generates a specific result, and that result can be
used to draw conclusions and make forecasts.
1 + Inflation Rate
• In budgeting, it is the only way to compute fixed and variable portions
of costs that contain both fixed and variable components (i.e., mixed
costs).
Shortcomings and limitations of regression analysis:
• Regression analysis requires historical data. If historical data is not
available, regression analysis cannot be used.
• Even when historical data is available, if there has been a significant
change in conditions, the use of past data is questionable for predicting
the future.
• In causal forecasting, the usefulness of the forecast depends upon
the choice of the independent variable. An inappropriate choice can
lead to misleading results.
• Statistical relations developed using regression analysis are valid only
for the range of data in the sample.
Learning curves describe the fact that the more experience an individual Nominal rate = [(Real rate + 1) * (Inflation rate + 1)] - 1
has with something, the more efficient he or she becomes in doing
that task.
The concept of learning curves means that higher costs per unit should
be expected early in production as part of start-up costs, and the costs
per unit should be expected to decline over time, up to a point. There is a
limit as to how fast or efficiently something can be done, and so the
learning curve is not something that continues indefinitely.
Part 1 CMA – Section D: Quantitative Methods Part 1 CMA – Section D: Quantitative Methods
©
HOCK international - 2004 17 ©
HOCK international - 2004 18
Part 1 CMA – Section D: Quantitative Methods Part 1 CMA – Section D: Quantitative Methods
©
HOCK international - 2004 19 ©
HOCK international - 2004 20
The incremental unit-time learning model assumes that each time The cumulative average-time learning model assumes that each
the cumulative quantity of units produced doubles, there will be a time the cumulative quantity of units produced doubles, there will be a
constant percentage of decline in the time needed to produce the constant percentage of decline in the average time per unit required
last unit. for the entire (cumulative) amount produced.
10 + 8 = 18 hours. After 4 units have been built, the total time required to build 4 units
will be 10 * (2 * .80)2, or 25.6 hours. Thus, the first unit took 10
hours; the second unit took 6 hours, and the third and fourth units
The average time per unit will be 18 ÷ 2 = 9 hours. took a total of 9.6 hours, or only 4.8 hours each.
There are three limitations and problems associated with learning The benefits of learning curve analysis are that it can be used in
curve analysis: making decisions such as:
1) It is appropriate for labor-intensive operations involving repetitive 1) Make or buy decisions, to analyze the cost to make.
tasks where repeated trials improve performance. However, if the 2) Life-cycle costing, in calculating the cost of a contract, leading to
process uses robotics and computer controls, there is little repetitive better bidding.
labor and little opportunity for learning to take place. 3) In cost-volume-profit analysis, to determine a more accurate
2) The learning rate is assumed to be constant. However, in actuality, breakeven point.
the decline in labor time might not be constant but instead 4) In development of standard costs, to adjust labor costs in
may vary. recognition of the fact that learning causes standard costs to
change over time.
3) The reliability of the learning curve calculation can be jeopardized
5) In capital budgeting, to project costs more accurately over the
because an observed change in productivity might actually be
associated with factors other than learning. It might be due to a life of the capital investment.
change in the labor mix, a change in the product mix, or other 6) In development of production plans and labor budgets.
factors. If so, a learning model that uses historical data would produce 7) In evaluation of management, to recognize that higher costs will
inaccurate estimates of labor time and cost. occur in the early phase of a product’s life cycle.
Part 1 CMA – Section D: Quantitative Methods Part 1 CMA – Section D: Quantitative Methods
©
HOCK international - 2004 21 ©
HOCK international - 2004 22
Part 1 CMA – Section D: Quantitative Methods Part 1 CMA – Section D: Quantitative Methods
©
HOCK international - 2004 23 ©
HOCK international - 2004 24
If the occurrence of one event means that another event cannot occur, Probability provides a means of measuring numerically how likely it is
the two events are said to be mutually exclusive. that an event will occur. It is a numerical measurement of uncertainty.
Probability is important in decision-making because it enables us to quantify
and analyze uncertainties.
The sum of the probabilities of mutually exclusive events is equal to 1.
Unless the probability is derived from data that results from actual
observations (objective), most probability has strong subjective elements.
Subjective probability is a numerical measure of the belief of an individual
in the occurrence or nonoccurrence of an event.
The joint probability of two events occurring is the probability that they
will both occur.
Joint probability may describe the probability of two events occurring The conditional probability of two events is the probability that one
when one is dependent upon the other’s occurring (conditional probability). will occur once it is known that the other has already occurred.
Or it may describe the probability of two events occurring that are
independent of each other, i.e., the occurrence of nonoccurrence of one
event does not change the probability of the occurrence of the other
event.
The probability of two independent events occurring is the independent
probability of the first event multiplied by the independent probability of
the second event.
If the second event is conditioned upon the first event occurring, the
probability that they both will occur is the probability of the first event
multiplied by the conditional probability of the second event.
The probability that either one or both of two events occurring is the
sum of their individual probabilities minus their joint probability.
Part 1 CMA – Section D: Quantitative Methods Part 1 CMA – Section D: Quantitative Methods
What Are the Two Basic What Are the Three Methods
Requirements of Probability? Used to Assign Probabilities?
©
HOCK international - 2004 25 ©
HOCK international - 2004 26
Part 1 CMA – Section D: Quantitative Methods Part 1 CMA – Section D: Quantitative Methods
©
HOCK international - 2004 27 ©
HOCK international - 2004 28
The three methods used to assign probabilities are: The two basic requirements of probability are:
1) Classical method. This assumes that every possible outcome has 1) The probability values assigned to each of the possible outcomes
an equal probability of occurring. The classical method is seldom must be between 0 and 1, and
used in situations of business uncertainty.
2) The probable values assigned to all of the possible outcomes must
2) Objective, or Relative Frequency, method. This is used when total 1.
factual information is available that can be used to determine the
probability of something occurring. The information may come from
sample data or any other reliable source.
3) Subjective method. This is used when factual information is not
available and the possible outcomes are not equally likely. Using
whatever data is available and our own experience and intuition,
we assign a probability that expresses our degree of belief that
the outcome will occur.
The expected value is the mean or average of the values of the A discrete random variable is a random variable that can take on
random variable. When we are working with probabilities, it is a weighted the value of any integer, i.e., any whole number, such as the number of
average of all the possible values of the random variable, people coming into a store, or the number of computers sold in a
with the probabilities for each of the values used as the weights. day’s time.
©
HOCK international - 2004 29 ©
HOCK international - 2004 30
Part 1 CMA – Section D: Quantitative Methods Part 1 CMA – Section D: Quantitative Methods
©
HOCK international - 2004 31 ©
HOCK international - 2004 32
The standard deviation of a probability distribution also gives us an The variance of a probability distribution gives us an idea of the
idea of the variability of the possible values for the random variable. variability of the possible values for the random variable. The variance
The standard deviation is the positive square root of the variance. is the sum of the squares of all the differences (deviations) from the
mean (or average). This difference from the mean of each result tells
us how far any particular measurement is from its expected value.
In a normal distribution, one standard deviation represents a
probability of about 68% that an actual observation will fall within the
interval of the amount of the standard deviation on either side of
the mean.
A probability distribution can be developed for a discrete random A continuous random variable is a random variable that can take on
variable by observing historical data. any value within an interval or a collection of intervals. It can be any
value whatsoever and can take on a fractional value.
Since continuous random variables can take on any value at all
within a specified interval, we cannot develop a probability distribution
for them by observing historical data. The probability of the random
variable having a value within the given interval is defined as the area
under a graph that is called a probability density function.
Part 1 CMA – Section D: Quantitative Methods Part 1 CMA – Section D: Quantitative Methods
©
HOCK international - 2004 33 ©
HOCK international - 2004 34
Part 1 CMA – Section D: Quantitative Methods Part 1 CMA – Section D: Quantitative Methods
©
HOCK international - 2004 35 ©
HOCK international - 2004 36
The normal probability distribution is the most important distribution. For a binomial distribution only, we can calculate the expected
The graph of a normal probability distribution has the form of a value of the distribution by multiplying the number of trials by the
symmetrical bell-shaped curve that is centered on its mean. The mean probability of success on any one trial.
and the standard deviation of the distribution explain it.
A larger mean causes the top of the curve to be higher. And a larger
standard deviation causes the curve to be flatter and broader because
the dispersion of the data about the mean is greater.
Risk in investing is the possibility that an investment’s actual return A standard normal distribution is a normal distribution that has a mean
will differ from its expected return. Risk for a security can be measured of 0 and a standard deviation of 1. Values in a standard normal distribution
by the variability of its historical returns or the dispersion of its table are available to use in calculating the area under the curve between
historical returns around their average, or mean, return. Thus, risk is the mean and any positive value along the graph. Therefore, the area
measured by variance and standard deviation. under the curve can be found by first converting it to a standard normal
distribution and then using the standard normal distribution table.
Uncertainty is risk that cannot be measured. If there is no historical
data to use for developing information to estimate probability and thus The area under the normal curve between the mean and one standard
expected return, we are in the position of decision-making under a deviation is approximately 68% of the total area.
condition of uncertainty. When this is the case, the probability The area under the normal curve between the mean and two standard
distribution of possible returns must be determined subjectively. deviations is approximately 95% of the total area.
The area under the normal curve between the mean and three standard
deviations is approximately 99.7% of the total area.
Part 1 CMA – Section D: Quantitative Methods Part 1 CMA – Section D: Quantitative Methods
©
HOCK international - 2004 37 ©
HOCK international - 2004 38
Part 1 CMA – Section D: Quantitative Methods Part 1 CMA – Section D: Quantitative Methods
©
HOCK international - 2004 39 ©
HOCK international - 2004 40
The coefficient of variation is a measure of risk per unit of expected For decisions involving risk and uncertainty, we use expected value
return. It compares the amount of the variation in expected returns with or expected return to express the most likely result of our decision.
the amount of the expected return. And we use the standard deviation of the probability distribution of
the potential returns as a measurement of the risk associated with the
decision.
Using the expected return expressed as a rate of return, the coefficient
of variation is: By expressing differences from the expected return in terms of standard
deviation, we can state the probability that the actual return will be
greater than or less than the expected return. The greater the standard
deviation, the greater the potential for great loss or great gain.
Coefficient of variation (CV) = ó
Expected Return
The median is the halfway value if raw data is arranged in numerical The mean is the arithmetic average of a set of numbers. It may also be a
order from lowest to highest. Thus, half the values are smaller than weighted average.
the median and half of the values are larger than the median.
The mean of a sample is often represented with a bar over the letter for
EXAMPLE: For a set of numbers: 1, 1, 3, 6, 7 the variable. The mean of a population is often represented by the
Greek letter mu (µ).
The median is 3 (half of the values are above it and half are below).
What Do Measures of
Measures of Central Tendency: Dispersion, Such as
What Is the Mode? Variance and Standard
Deviation, Indicate?
©
HOCK international - 2004 41 ©
HOCK international - 2004 42
Part 1 CMA – Section D: Quantitative Methods Part 1 CMA – Section D: Quantitative Methods
Measures of Dispersion:
Measures of Dispersion: What Is the
What Is the Variance? Standard Deviation?
©
HOCK international - 2004 43 ©
HOCK international - 2004 44
Measures of dispersion such as variance and standard deviation The mode of a group of data is the most frequently occurring value
indicate the amount of variation within a set of numbers with respect in the data. If all values are unique (different from each other), no
to the mean of those numbers. mode exists.
The variance is the average of the squared deviations from the mean.
The standard deviation is the square root of the variance. It is a measure
The variance of a population is represented by the lowercase Greek letter
of how close together all of the items in the population are to the
sigma squared. The formula for the variance of a set is:
population’s mean. A population in which all of the values are very close
to the mean will have a low standard deviation.
N
( x i - µ )2
The formula to calculate the standard deviation is the square root of the ó 2= Σ
i =1 N
formula for the variance.
©
HOCK international - 2004 45 ©
HOCK international - 2004 46
Part 1 CMA – Section D: Quantitative Methods Part 1 CMA – Section D: Quantitative Methods
©
HOCK international - 2004 47 ©
HOCK international - 2004 48
The benefits of decision trees are: A decision tree is a means of determining the best course of action
under a condition of risk when there are several possible decision choices.
1) They are helpful when there is a series of conditional choices to
be made. Decision trees are used with probabilities to determine the expected value
of the payoff of a project that may involve making several decisions.
2) They show the impact of time on decisions.
3) They can model uncertainty. For each sequence of events a payoff is computed, and for each decision
alternative an expected monetary value (EMV) is found by summing the
4) They produce quantitative results. products of the payoffs and their respective probabilities. The decision
5) They are flexible, examining the effects of predictors one at a time. leading to the highest EMV is optimal.
Decision trees are used for capital budgeting, allocating budget monies
among possible projects or for possible R&D activities. They are also used
for decisions such as developing and deciding on new marketing strategies.
The expected value of perfect information is the difference The shortcomings of decision trees are:
between the expected payoff we could receive if we had perfect
information about the future, and the expected payoff we can receive 1) All decision factors must be expressed quantitatively. Qualitative factors
by using our best analytical decision-making skills, but without perfect are difficult, if not impossible, to express and utilize.
information about the future.
2) Decision trees can be a challenge to develop in a group setting because
of the difficulty of reaching agreement on event probabilities due to
their frequently subjective nature.
4) All data developed from decision trees must be subjected to the good
judgment of the decision-maker(s).
Part 1 CMA – Section D: Quantitative Methods Part 1 CMA – Section D: Quantitative Methods
©
HOCK international - 2004 49 ©
HOCK international - 2004 50
Part 1 CMA – Section D: Quantitative Methods Part 1 CMA – Section D: Quantitative Methods
©
HOCK international - 2004 51 ©
HOCK international - 2004 52
A linear programming problem involves decisions that must be made, Linear programming is a method of problem-solving that is used to
such as the quantity of each product to produce. These decisions are either maximize or minimize some quantity, while at the same time
represented by decision variables. accomplishing this in the presence of constraints or restrictions.
For instance, if we are determining the optimal quantities to produce The restriction may be something such as limited quantities of labor or
of two products, we will use decision variables in our objective equation materials. The maximizing or minimizing must be done without violating
and constraint inequalities. If X represents the quantity of the first any of the constraints.
product and Y represents the quantity of the second product, then X Typically, linear programming is used to maximize the total contribution
and Y are the decision variables. margin of a mix of products under conditions of multiple constraints.
It is also used for other goals, such as profit levels, total revenues,
total costs, pollution levels and percent return on an investment.
The two types of constraints in a linear programming model are: Constraints in linear programming represent conditions that must be
• Structural constraints, which are used to express things such as satisfied. Constraints might be limited raw materials, equipment time,
resource limitations, and labor hours or even limited demand for the products.
• Nonnegativity constraints, which are included to ensure that Constraints in a linear programming problem are usually represented by
decision variables will be only positive numbers. inequalities (i.e., < or <= and > or => types).
©
HOCK international - 2004 53 ©
HOCK international - 2004 54
Part 1 CMA – Section D: Quantitative Methods Part 1 CMA – Section D: Quantitative Methods
©
HOCK international - 2004 55 ©
HOCK international - 2004 56
If a particular constraint has an unused portion, that constraint is called The objective function in linear programming is the equation that
a nonbinding constraint. expresses the objective that we are trying to accomplish. The optimal
solution is represented by Z in the objective function.
The slack variable in linear programming is a variable added to an inequality
in a computer program to represent the unused (slack) capacity associated
Of all of the possible combinations that will not violate the constraints,
with the constraint. The slack variable represents the difference between
the objective function attempts to find the one feasible solution that
the right-hand side and the left-hand side of a = constraint.
will either maximize or minimize Z, whichever one we are trying to do.
Slack variables are usually represented by S1, S 2, S3, etc.
An example of an objective function is:
z = 130x + 175y
The shadow price is a term used when sensitivity analysis is done in The dual price is a term used when sensitivity analysis is done in
conjunction with linear programming. By changing a coefficient in the conjunction with linear programming. By changing a coefficient in the
objective function or by changing the right-hand value of one of the objective function, or by changing the right-hand value of one of the
constraint functions, we can see how the optimal solution to the constraint functions, we can see how the optimal solution to the
objective function would change. objective function would change.
©
HOCK international - 2004 57 ©
HOCK international - 2004 58
Part 1 CMA – Section D: Quantitative Methods Part 1 CMA – Section D: Quantitative Methods
©
HOCK international - 2004 59 ©
HOCK international - 2004 60
The algebraic solution to a linear programming problem can be used to If there is a linear programming problem with only two variables, it can
find the one unique solution to the problem only if the number of be solved graphically by:
constraint equations is equal to the number of variables in the 1) Changing the constraint inequalities to equalities and plotting the
constraint equations. If there are more variables than constraint resulting equations on a graph.
equations, no unique solution exists.
2) Identifying the side of each line that represents the feasible area of
To find a unique solution when the number of constraint equations is the solutions for that constraint inequality.
equal to the number of variables:
3) Defining the feasible solution, which is within the area to the left of
1) Change the constraint inequalities to equations; the two constraint equations. The optimal solution will be at one of
2) Solve the constraint equations simultaneously to find the unique the corners of the feasible solution.
solution; 4) Substituting the coordinates for each of the corner feasible solutions
3) Then substitute the values you get for the variables into the objective into the objective function to see which set of coordinates maximizes
function. (or minimizes, if that is the goal) the objective function.
The limitations of linear programming are: The benefits of linear programming are in its versatility and flexibility
for multiple uses and the fact that it can be used with sensitivity analysis.
• The decision problem must be expressed in linear functions, which
requires the assumption that all costs are either variable or fixed • It can be used not only for contribution margin optimization, but also in
costs. If the relationships involved are not linear, then linear advertising for media selection to get the maximum amount of advertising
programming will not provide an answer. exposure given the types and cost of the available media.
• Since the very best possible solution is sought, minor variances • In production management, linear programming can be used to determine
between assumed coefficient values and actual coefficient values how many and which parts a company should manufacture and how
can have a drastic impact upon the resulting solution. many it should purchase.
©
HOCK international - 2004 61 ©
HOCK international - 2004 62
Part 1 CMA – Section D: Quantitative Methods Part 1 CMA – Section D: Quantitative Methods
©
HOCK international - 2004 63 ©
HOCK international - 2004 64
Integer linear programming problems have an additional requirement Transportation problems represent a special class of linear programming
that one or more of the variables must be an integer. Usually, the simplest problems that are called network flow problems.
way to solve these is by simply rounding the linear programming solution.
They are used in planning for distribution of goods from several different
However, if one or more of the integer variables must equal either 0 or 1, source locations, such as factories, to several different locations where
rounding is not acceptable because it would lead to too great a rounding they are needed, such as warehouses or wholesalers.
error.
The transportation costs of each item are used as the coefficients in the
Sensitivity analysis is usually more important in integer linear programming objective function, and the objective is to minimize the cost of shipping
because a one-unit change in one of the coefficients can cause a relatively by determining the optimal routes to use and the optimal quantity to ship
large change in the optimal solution. via each route in order to minimize total transportation cost.
The constraints are used to express the maximum supply available from
each source location and the demand of each destination.
©
HOCK international - 2004 65 ©
HOCK international - 2004 66
Part 1 CMA – Section D: Quantitative Methods Part 1 CMA – Section D: Quantitative Methods
Uncertainty in PERT/CPM
What Is the Critical Path in Activity Times: What Is a
a PERT/CPM Network? Probabilistic Technique?
©
HOCK international - 2004 67 ©
HOCK international - 2004 68
The most important concept of PERT/CPM is that one group of Project scheduling is the process of planning, managing and controlling
activities controls the entire project, because it is that set of large projects composed of many different jobs that are performed by
activities that will take the longest time to complete. many different departments and people.
A probabilistic technique, also called a stochastic technique, is a A path through the network is a series of connected nodes (representing
technique that allows for uncertainty. activities) that goes all the way from the beginning of the project to
the end. A network may have many paths, and all of them must be
This is in contrast to a deterministic technique, which does not
completed in order to complete the project.
allow for uncertainty.
The critical path requires the most time to complete. If activities on
that path are delayed for any reason, the entire project will be delayed.
Activities on the critical path are called critical activities.
©
HOCK international - 2004 69 ©
HOCK international - 2004 70
Part 1 CMA – Section D: Quantitative Methods Part 1 CMA – Section D: Quantitative Methods
©
HOCK international - 2004 71 ©
HOCK international - 2004 72
Using the assumption that one standard deviation is approximately 1/6 th To determine the Expected Completion Time for an activity in a
of the difference between the most extreme values of a probability PERT/CPM diagram when times are uncertain:
distribution, we determine the standard deviation (ó) of an individual
activity as follows: • Make three time estimates: its optimistic time, its most
probable time and its pessimistic time.
ó = Pessimistic time - Optimistic time
• Using those three time estimates, use the following formula to
6 calculate the Expected Value of the Completion Time:
Benefits of PERT/CPM analysis are: Crashing an activity refers to putting additional resources to work on a
• It forces managers to plan projects in intricate detail. specific activity in order to shorten the time to complete the project.
• It can be used for scheduling. Crashing is done on activities that are on the critical path. However, if we
crash these activities too much, then it can make another path into a
• It can be used to assign existing resources to a project in the most
critical path and waste some of our additional resources.
effective manner.
• It can be used to calculate costs to shorten the time required for the CPM brings the concept of the cost-time tradeoff to the network analysis.
project. The cost of the additional resources required to shorten an activity’s
completion time is balanced against the benefit received from
• Sensitivity analysis can be used with PERT/CPM to determine the
the shortening.
probability of finishing a project on time.
The activity or activities with the lowest cost in relation to the benefit
received from crashing them are the activities to be crashed.
Part 1 CMA – Section D: Quantitative Methods Part 1 CMA – Section D: Quantitative Methods
©
HOCK international - 2004 73 ©
HOCK international - 2004 74
Part 1 CMA – Section D: Quantitative Methods Part 1 CMA – Section D: Quantitative Methods
©
HOCK international - 2004 75 ©
HOCK international - 2004 76
Limitations of PERT/CPM analysis are:
Simulation analysis is the process of learning about a real situation • PERT is extremely complicated. When costs are introduced through
or system by experimenting with a model that represents the real CPM analysis, scheduling complexity is increased.
• CPM doesn’t deal with influences of indirect costs and contractual
situation or system. Given certain values for inputs, the simulation
incentives but assumes that time and costs are linearly related, which
model uses mathematical expressions and logical relationships to they may not be.
compute the value of the outputs. • PERT/CPM can lead to overly optimistic estimates.
• Looking only at the critical path can be misleading, because paths that
are near-critical and that have large variances may become critical.
• The three estimates may be only guesses and thus not necessarily
any better than one guess. The actual activity times required may be
quite different from their expected values.
• PERT/CPM considers the various activities to be independent and the
expected length of a project is the sum of the activities’ separate
expected lengths. This assumption may not be correct, because many
activities have dependencies.
• If PERT/CPM is used to shorten the time required for a project by
eliminating or substituting activities, it can result in degrading of
requirements and poor quality of work.
A Monte Carlo simulation is a specific type of simulation. It allows The two types of inputs that a simulation model has are:
assessment of the probabilities of various scenarios by generating random
values for the probabilistic inputs. Each of these values is based on 1) Controllable inputs, or inputs that are selected by the
the probability distribution for each input. These values are then used decision-maker; and
to generate multiple, possible scenarios. Enough trials are conducted
2) Probabilistic inputs, or inputs that are subject to uncertainty
(hundreds or thousands), using different values for the probabilistic
and described by probability distributions.
inputs, to determine a probability distribution for the resulting scenario,
which is the output.
©
HOCK international - 2004 77 ©
HOCK international - 2004 78
Part 1 CMA – Section D: Quantitative Methods Part 1 CMA – Section D: Quantitative Methods
©
HOCK international - 2004 79 ©
HOCK international - 2004 80
The limitations of simulation analysis include: The benefits of simulation analysis include: