Project Sales or Production Levels Using the Rolling Average © 20111.

Project Sales or Production Project Sales or Production Levels Using the Rolling AverageLevels Using the Rolling Average

© 2011 1

What if?What if?

You planned for 10 but…

© 2011

Terminal Learning ObjectiveTerminal Learning Objective• Task: Project Sales or Production Levels Using the

Rolling Average• Condition: You are training to become an ACE

with access to ICAM course handouts, readings, and spreadsheet tools and awareness of Operational Environment (OE)/Contemporary Operational Environment (COE) variables and actors

• Standard: with at least 80% accuracy• Demonstrate understanding of Trend Projection

concepts

© 2011 3

Importance of DemandImportance of Demand

• We have seen how demand drives cost• Flexible forecasting

• Assumptions about probabilities may not yield useful information• “Precisely wrong”

• Examining trends gives another perspective on demand

© 2011 4

Predicting the FuturePredicting the Future

© 2011 5

What is Trend Projection?What is Trend Projection?

• Uses historical data about past demand to make estimates of future demand

• Relies on systematic methodologies and assumptions

• Cannot predict the future or anticipate catastrophic events

© 2011 6

Three MethodsThree Methods

• Regression• Represents a straight line with the least

squared error from actual

• Rolling average• Uses average of prior period demand to predict

future period demand

• Planning factors• Assumes a relationship between a current

value and future demand© 2011 7

Regression AnalysisRegression Analysis

• Plots a linear relationship between multiple data points

• Minimizes the “squared errors”• Square difference between mean and actual to

eliminate negative values

• Uses the format y = mx + b where:

© 20118

Regression ResultsRegression Results

• Very predictable • The ascending series is y = x + 4 and we can predict that

the 7th period would need 11 burgers• The descending series is y = -x + 17 and we can predict that

the 7th period would need 10

© 2011 9

Regression ExerciseRegression Exercise

• Use spreadsheet to predict the 8th, 9th, and 10th event burger demand if the first six demands were:• 8 10 9 12 13 15

© 2011 10

Spreadsheet ExerciseSpreadsheet Exercise

© 2011 11

The spreadsheet returns the equation:y = 1.4x + 6.2667Enter the values in the equation to project future demand

Demand for:Period 8 = 17Period 9 = 19Period 10 = 20

Example: Using Regression to Estimate Example: Using Regression to Estimate Fixed and Variable CostsFixed and Variable Costs

• Consider four quarters of data

• Regression returns y = 2.2x +13.7

13

Q1 Q2 Q3 Q4

Units 5 6 7 8

Total Cost 25 27 28 32

Fixed cost is 13.7

Variable cost is 2.2 per unit

Total cost is 13.7 + 2.2*units


© 2011 14

Notice that four very different sets of data all have very similar regression lines

The x-axis in these graphs represents time periods in series

Regression Strengths and WeaknessesRegression Strengths and Weaknesses

• Can be calculated very precisely• But cumbersome to do by hand(use spreadsheet!)• May be precisely wrong

• Can be used to identify trends• But by definition cannot predict downturns or

upturns• Assumes relationship is linear and will remain

linear

© 2011 15

Learning CheckLearning Check

• In the context of trend projection, what does the regression line represent?

• What is the main weakness of regression in trend projection?

© 2011 16

Rolling AverageRolling Average

• Uses average of prior periods to predict future periods

• Evens out highs and lows by using a number of periods

• Key assumption for predictions:• Assumes that the average will be maintained• Example: Average of Periods 2, 3 & 4 will equal

average of periods 1, 2 & 3

© 2011 17

Rolling Average CalculationRolling Average Calculation

• The demand for our last twelve periods has been:

• Task: Calculate the 3-month rolling average for periods 3-12

© 2011 18


• The 3-month rolling average is the average value for the most recent 3 months

Per1 + Per2 + Per33

• Add the most recent period to the calculation and drop the oldest

Per2 + Per3 + Per43

© 2011 19


© 2011 20

Period1 not enough data2 not enough data3 (6 + 8 + 4)/3 = 6.04 (8 + 4 + 3)/3 = 5.05 (4 + 3 + 7)/3 = 4.76 (3 + 7 + 5)/3 = 5.0

Period7 (7 + 5 + 6)/3 = 6.08 (5 + 6 + 8)/3 = 6.39 (6 + 8 + 3)/3 = 5.710 (8 + 3 + 6)/3 = 5.711 (3 + 6 + 4)/3 = 4.312 (6 + 4 + 5)/3 = 5.0


© 2011 21

Period7 (7 + 5 + 6)/3 = 6.08 (5 + 6 + 8)/3 = 6.39 (6 + 8 + 3)/3 = 5.710 (8 + 3 + 6)/3 = 5.711 (3 + 6 + 4)/3 = 4.312 (6 + 4 + 5)/3 = 5.0


© 2011 22

Period7 (7 + 5 + 6)/3 = 6.08 (5 + 6 + 8)/3 = 6.39 (6 + 8 + 3)/3 = 5.710 (8 + 3 + 6)/3 = 5.711 (3 + 6 + 4)/3 = 4.312 (6 + 4 + 5)/3 = 5.0


© 2011 23

Period7 (7 + 5 + 6)/3 = 6.08 (5 + 6 + 8)/3 = 6.39 (6 + 8 + 3)/3 = 5.710 (8 + 3 + 6)/3 = 5.711 (3 + 6 + 4)/3 = 4.312 (6 + 4 + 5)/3 = 5.0


© 2011 24

Period1 not enough data2 not enough data3 (6 + 8 + 4)/3 = 6.04 (8 + 4 + 3)/3 = 5.05 (4 + 3 + 7)/3 = 4.76 (3 + 7 + 5)/3 = 5.0

Period7 (7 + 5 + 6)/3 = 6.08 (5 + 6 + 8)/3 = 6.39 (6 + 8 + 3)/3 = 5.710 (8 + 3 + 6)/3 = 5.711 (3 + 6 + 4)/3 = 4.312 (6 + 4 + 5)/3 = 5.0

Graph of Rolling AverageGraph of Rolling Average

© 2011 25

This is a time series. X-axis represents sequential time periods

Graph of Rolling AverageGraph of Rolling Average

© 2011 26


Rolling Average vs. RegressionRolling Average vs. Regression

© 2011 27


Using Rolling Average to Project Future Using Rolling Average to Project Future DemandDemand

• Assume that the previous rolling average will be maintained

• Our forecast for period 13 will assume a rolling average of 5, same as period 12

(Per11 + Per12 + Per13)/3 = 5

© 2011 28


• Plug in the known values and solve the equation:

(Per11 + Per12 + Per13)/3 = 5(4 + 5 + Per13)/3 = 5

3 * (4 + 5 + Per13)/3 = 5 * 39 + Per13 = 15

Per13 = 6

© 2011 29


• Plug in the known values and solve the equation:

(Per11 + Per12 + Per13)/3 = 5(4 + 5 + Per13)/3 = 5

3 * (4 + 5 + Per13)/3 = 5 * 39 + Per13 = 15

Per13 = 6

© 2011 30

What would regression analysis project?

Which is “right”?

Rolling Average vs. RegressionRolling Average vs. Regression

© 2011 31

This is a time series. X-axis represents sequential time periods13

3 month rolling average suggests an inflection point has

changed the trend

Regression picks up the long term downward trend,

predicting another decrease

Rolling Average Strengths and WeaknessesRolling Average Strengths and Weaknesses

• Can be calculated very precisely• But may be precisely wrong

• Simple to calculate• The main strength of rolling averages is that they

dampen the effect of short term changes• This helps decision makers avoid knee jerk responses to

changes in demand that may not be significant• Decision makers are often looking for inflection points• An inflection point in a six month rolling average carries a

lot of weight

© 2011 32


• What would be the equation for a six-month rolling average calculation?

• What is the primary assumption when using rolling average to project future demand?

© 2011 33

Planning FactorsPlanning Factors

• Assume some cause and effect relationship• If we suspect that demand for education

counseling decreases when a unit deploys• We could study the history of that relationship

and determine a planning factor (or ratio) of sessions per soldier as “a”

• We could then use that factor to plan for the drop in session demand when X soldiers deploy as

• New demand = a*X

© 2011 34

Planning Factor ExamplePlanning Factor Example

• Given the recent history determine the planning factor relating sessions and soldiers

• Use that factor to predict sessions as population goes to• 8000• 7000• 6000

© 2011 35

Planning Factor ExamplePlanning Factor Example

• Given the recent history determine the planning factor relating sessions and soldiers

• Use that factor to predict sessions as population goes to• 8000 * .032 = 256• 7000 * .032 = 224• 6000 * .032 = 192

© 2011 36

Total = 1994 623651994/62365 = .032 or 3.2%

Leading IndicatorsLeading Indicators

• Leading indicators are similar to planning factors with a couple differences• Leading indicators often have a weaker cause and

effect relationship• Changes in consumer confidence index may

foreshadow an increase in sales at the post exchange

• There is a period of time before the effect is seen (i.e. that’s why they are called leading indicators)

© 2011 37

Project Sales or Production Levels Using the Rolling Average © 20111.

Documents

Transcript of Project Sales or Production Levels Using the Rolling Average © 20111.