DFMD 3513 Chapter-6 Forecasting | Correlation And Dependence | Forecasting

Ir.

Haery Sihombing/IP Sihombing/IP
Pensyarah Fakulti Kejuruteraan Pembuatan

Universiti Teknologi Malaysia Melaka

What is Forecasting?
Process of predicting a future event Underlying basis of all business decisions
Production Inventory Personnel Facilities

6
FORECASTING

??

Forecasting
Predicting the Future Qualitative forecast methods
subjective

Forecasting and Supply Chain Management
Accurate forecasting determines how much inventory a company must keep at various points along its supply chain Continuous replenishment
supplier and customer share continuously updated data typically managed by the supplier reduces inventory for the company speeds customer delivery

Quantitative forecast methods
based on mathematical formulas

Variations of continuous replenishment
quick response JIT (just-in-time) (just- inVMI (vendor-managed inventory) (vendorstockless inventory

Forecasting and TQM
Accurate forecasting customer demand is a key to providing good quality service Continuous replenishment and JIT complement TQM
eliminates the need for buffer inventory, which, in turn, reduces both waste and inventory costs, a primary goal of TQM smoothes process flow with no defective items meets expectations about on-time delivery, which is onperceived as good-quality service good-

Types of Forecasting Methods
Depend on
time frame demand behavior causes of behavior

Time Frame
Indicates how far into the future is forecast
Short- to mid-range forecast Shortmidtypically encompasses the immediate future daily up to two years

Forecasting Time Horizons
Short-range forecast ShortUp to 1 year, generally less than 3 months Purchasing, job scheduling, workforce levels, job assignments, production levels

Medium-range forecast Medium3 months to 3 years Sales and production planning, budgeting

Long-range forecast Longusually encompasses a period of time longer than two years

Long-range forecast Long3+ years New product planning, facility location, research and development

Distinguishing Differences
Medium/long range forecasts deal with more comprehensive issues and support management decisions regarding planning and products, plants and processes Short-term forecasting usually employs Shortdifferent methodologies than longer-term longerforecasting Short-term forecasts tend to be more Shortaccurate than longer-term forecasts longer-

Influence of Product Life Cycle
Introduction – Growth – Maturity – Decline Introduction and growth require longer forecasts than maturity and decline As product passes through life cycle, forecasts are useful in projecting
Staffing levels Inventory levels Factory capacity

Product Life Cycle
Introduction Company Strategy/Issues
Best period to increase market share R&D engineering is critical

Product Life Cycle
Decline
Cost control critical

Growth
Practical to change price or quality image Strengthen niche

Maturity
Poor time to change image, price, or quality Competitive costs become critical Defend market position

Introduction
Product design and development critical Frequent product and process design changes Short production runs High production costs Limited models Attention to quality

Growth
Forecasting critical Product and process reliability Competitive product improvements and options Increase capacity Shift toward product focus Enhance distribution

Maturity
Standardization Less rapid product changes – more minor changes Optimum capacity Increasing stability of process Long production runs Product improvement and cost cutting

Decline
Little product differentiation Cost minimization Overcapacity in the industry Prune line to eliminate items not returning good margin Reduce capacity

CD-ROM CDInternet Sales Color printers Drive-through Driverestaurants

Fax machines

Flat-screen Flatmonitors

DVD

3 1/2” 1/2” Floppy disks

OM Strategy/Issues

Figure 2.5

Figure 2.5

Types of Forecasts
Economic forecasts Address business cycle – inflation rate, money supply, housing starts, etc. Technological forecasts Predict rate of technological progress Impacts development of new products Demand forecasts Predict sales of existing product

Strategic Importance of Forecasting
Human Resources – Hiring, training, laying off workers Capacity – Capacity shortages can result in undependable delivery, loss of customers, loss of market share Supply-Chain Management – Good supplier Supplyrelations and price advance

Seven Steps in Forecasting
1. 2. 3.

The Realities!
Forecasts are seldom perfect Most techniques assume an underlying stability in the system Product family and aggregated forecasts are more accurate than individual product forecasts

4. 5. 6. 7.

Determine the use of the forecast Select the items to be forecasted Determine the time horizon of the forecast Select the forecasting model(s) Gather the data Make the forecast Validate and implement results

Forecasting Approaches
Qualitative Methods
Used when situation is vague and little data exist
New products New technology

Forecasting Approaches
Quantitative Methods
Used when situation is ‘stable’ and stable’ historical data exist
Existing products Current technology

Involves intuition, experience
e.g., forecasting sales on Internet

Involves mathematical techniques
e.g., forecasting sales of color televisions

Qualitative Methods
o

Overview of Qualitative Methods
Jury of executive opinion
Pool opinions of high-level executives, sometimes highaugment by statistical models

o

Management, marketing, purchasing, and engineering are sources for internal qualitative forecasts Delphi method
involves soliciting forecasts about technological advances from experts

Delphi method
Panel of experts, queried iteratively

Sales force composite
Estimates from individual salespersons are reviewed for reasonableness, then aggregated

Consumer Market Survey
Ask the customer

Jury of Executive Opinion
Involves small group of high-level managers highGroup estimates demand by working together Combines managerial experience with statistical models Relatively quick ‘Group-think’ Group-think’ disadvantage

Sales Force Composite
Each salesperson projects his or her sales Combined at district and national levels Sales reps know customers’ wants customers’ Tends to be overly optimistic

Delphi Method
Iterative group process, continues until consensus is reached
Staff (Administering survey) Decision Makers (Evaluate responses and make decisions)

Consumer Market Survey
Ask customers about purchasing plans What consumers say, and what they actually do are often different Sometimes difficult to answer

3 types of participants
Decision makers Staff Respondents
Respondents (People who can make valuable judgments)

Overview of Quantitative Approaches
1. 2. 3.

Time Series Forecasting
Set of evenly spaced numerical data

4. 5.

Naive approach Moving averages Exponential smoothing Trend projection Linear regression

Time-Series TimeModels

Obtained by observing response variable at regular time periods

Forecast based only on past values
Associative Model

Assumes that factors influencing past and present will continue influence in future

Time Series Components
Demand for product or service

Components of Demand
Trend component Seasonal peaks

Trend

Cyclical

Actual demand Average demand over four years
| 3 Year | 4

Seasonal

Random

Random variation
| 1 | 2

Demand Behavior
Trend Random variations
movements in demand that do not follow a pattern a gradual, long-term up or down movement of longdemand

Forms of Forecast Movement
Demand Demand Random movement Time (a) Trend Time (b) Cycle Demand Time (c) Seasonal pattern Time (d) Trend with seasonal pattern

Cycle
an up-and-down repetitive movement in demand up- and-

Seasonal pattern
Demand

an up-and-down repetitive movement in demand up- andoccurring periodically

Trend Component
Persistent, overall upward or downward pattern Changes due to population, technology, age, culture, etc. Typically several years duration

Seasonal Component
Regular pattern of up and down fluctuations Due to weather, customs, etc. Occurs within a single year
Period Week Month Month Year Year Year Length Day Week Day Quarter Month Week Number of Seasons 7 4-4.5 28-31 4 12 52

Cyclical Component
Repeating up and down movements Affected by business cycle, political, and economic factors Multiple years duration Often causal or associative relationships
0 5 10 15 20

Random Component
Erratic, unsystematic, ‘residual’ residual’ fluctuations Due to random variation or unforeseen events Short duration and nonrepeating

M

T

W

T

F

Forecasting Methods
Qualitative
use management judgment, expertise, and opinion to predict future demand

Forecasting Process
1. Identify the purpose of forecast 2. Collect historical data 3. Plot data and identify patterns

Time series
statistical techniques that use historical demand data to predict future demand

6. Check forecast accuracy with one or more measures 7. Is accuracy of forecast acceptable?

5. Develop/compute forecast for period of historical data

4. Select a forecast model that seems appropriate for data

No

Regression methods
attempt to develop a mathematical relationship between demand and factors that cause its behavior

8b. Select new forecast model or adjust parameters of existing model

Yes
8a. Forecast over planning horizon 9. Adjust forecast based on additional qualitative information and insight 10. Monitor results and measure forecast accuracy

Time Series
Assume that what has occurred in the past will continue to occur in the future Relate the forecast to only one factor - time Include
moving average exponential smoothing linear trend line

Moving Average
Naive forecast
demand the current period is used as next period’s forecast period’

Simple moving average
stable demand with no pronounced behavioral patterns

Weighted moving average
weights are assigned to most recent data

Naive Approach
Assumes demand in next period is the same as demand in most recent period
e.g., If October sales were 90, then November sales will be 90
MONTH Jan Feb Mar Apr May June July Aug Sept Oct Nov

Naïve Approach
ORDERS PER MONTH 120 90 100 75 110 50 75 130 110 90 FORECAST 120 90 100 75 110 50 75 130 110 90

Sometimes cost effective and efficient

Simple Moving Average

Simple Moving Average
MA is a series of arithmetic means Used if little or no trend Used often for smoothing
Provides overall impression of data over time

n i=1

Di

MAn =
where

n

n =number of periods in the moving average Di = demand in period i

Moving average =

demand in previous n periods n

3-month Simple Moving Average
EXAMPLE

5-month Simple Moving Average
EXAMPLE

MONTH Jan Feb Mar Apr May June July Aug Sept Oct Nov

ORDERS PER MONTH 120 90 100 75 110 50 75 130 110 90 -

MOVING AVERAGE – – – 103.3 88.3 95.0 78.3 78.3 85.0 105.0 110.0

3

MA3 =

i=1

Di

MONTH Jan Feb Mar Apr May June July Aug Sept Oct Nov

ORDERS PER MONTH 120 90 100 75 110 50 75 130 110 90 -

MOVING AVERAGE – – – – – 99.0 85.0 82.0 88.0 95.0 91.0

5

3 90 + 110 + 130 3

MA5 =

i=1

Di

5

=

=

90 + 110 + 130+75+50 5 = 91 orders for Nov

= 110 orders for Nov

Smoothing Effects
150 – 125 – 100 – Orders 75 – 50 – 25 – 0– | Jan | Feb | Mar Actual 3-month 5-month

Weighted Moving Average
WMAn = Wi Di i=1
i=1

| | | | Apr May June July Month

| | Aug Sept

| Oct

| Nov

Adjusts moving average method to more closely reflect data fluctuations

where

Wi = the weight for period i,
between 0 and 100 percent

Wi = 1.00

Weighted Moving Average
Used when trend is present
Older data usually less important

Weighted Moving Average
Example 1
MONTH August September October November Forecast WEIGHT 17% 33% 50% DATA 130 110 90
3 i=1

Weights based on experience and intuition
Weighted moving average = (weight for period n) n) x (demand in period n) n) weights

WMA3 =

Wi D i

= (0.50)(90) + (0.33)(110) + (0.17)(130) = 103.4 orders

Example 2

Weights Applied 3 2 1 6

Period Last month Two months ago Three months ago Sum of weights

Potential Problems With Moving Average
Increasing n smooths the forecast but makes it less sensitive to changes Do not forecast trends well Require extensive historical data

Month
January February March April May June July

Actual Shed Sales
10 12 13 16 19 23 26

3-Month Weighted Moving Average

[(3 x 13) + (2 x 12) + (10)]/6 = 121/6 13) 12) (10)]/6 [(3 x 16) + (2 x 13) + (12)]/6 = 141/3 [(3 x 19) + (2 x 16) + (13)]/6 = 17 [(3 x 23) + (2 x 19) + (16)]/6 = 201/2

Moving Average And Weighted Moving Average
Weighted moving average

Exponential Smoothing
Averaging method Weights most recent data more strongly Reacts more to recent changes Widely used, accurate method

30 – Sales demand 25 – 20 – 15 – 10 – 5 – | J | F | M | A | M | J | J | A | S | O | N Actual sales Moving average

| D

Exponential Smoothing
Form of weighted moving average
Weights decline exponentially Most recent data weighted most

Exponential Smoothing (cont.)
Ft +1 =
where: Ft +1 = forecast for next period Dt = actual demand for present period Ft = previously determined forecast for present period = weighting factor, smoothing constant

Dt + (1 - )Ft

Requires smoothing constant ( )
Ranges from 0 to 1 Subjectively chosen

Involves little record keeping of past data

OR

Exponential Smoothing (cont.)
New forecast = last period’s forecast period’ + (last period’s actual demand period’ – last period’s forecast) period’ forecast) Ft = Ft – 1 + (At – 1 - Ft – 1)
where Ft = new forecast Ft – 1 = previous forecast = smoothing (or weighting) constant (0 1)

Effect of Smoothing Constant
0.0 1.0 = 0.20, then Ft +1 = 0.20 Dt + 0.80 Ft If If = 0, then Ft +1 = 0 Dt + 1 Ft 0 = Ft
Forecast does not reflect recent data

If

= 1, then Ft +1 = 1 Dt + 0 Ft = Dt
Forecast based only on most recent data

Exponential Smoothing ( =0.30)
PERIOD 1 2 3 4 5 6 7 8 9 10 11 12 MONTH Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec DEMAND 37 40 41 37 45 50 43 47 56 52 55 54 F2 = D1 + (1 - )F1 = (0.30)(37) + (0.70)(37) = 37 F3 = D2 + (1 - )F2 = (0.30)(40) + (0.70)(37) = 37.9 F13 = D12 + (1 - )F12 = (0.30)(54) + (0.70)(50.84) = 51.79

Exponential Smoothing (cont.)
PERIOD 1 2 3 4 5 6 7 8 9 10 11 12 13 MONTH Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec Jan DEMAND 37 40 41 37 45 50 43 47 56 52 55 54 – FORECAST, Ft + 1 ( = 0.3) ( = 0.5) – 37.00 37.90 38.83 38.28 40.29 43.20 43.14 44.30 47.81 49.06 50.84 51.79 – 37.00 38.50 39.75 38.37 41.68 45.84 44.42 45.71 50.85 51.42 53.21 53.61

Exponential Smoothing (cont.)
70 – 60 – 50 – Orders 40 – = 0.30 30 – Actual = 0.50

Adjusted Exponential Smoothing

where

AFt +1 = Ft +1 + Tt +1
T = an exponentially smoothed trend factor Tt +1 = (Ft +1 - Ft) + (1 - ) Tt Tt = the last period trend factor = a smoothing constant for trend

where
20 – 10 – 0– | 1 | 2 | 3 | 4 | 5 | 6 Month | 7 | 8 | 9 | 10 | 11 | 12 | 13

Adjusted Exponential Smoothing
( =0.30)
PERIOD 1 2 3 4 5 6 7 8 9 10 11 12 MONTH Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec DEMAND 37 40 41 37 45 50 43 47 56 52 55 54

Adjusted Exponential Smoothing:
Example
PERIOD 1 2 3 4 5 6 7 8 9 10 11 12 13 MONTH Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec Jan DEMAND 37 40 41 37 45 50 43 47 56 52 55 54 – FORECAST Ft +1 37.00 37.00 38.50 39.75 38.37 38.37 45.84 44.42 45.71 50.85 51.42 53.21 53.61 TREND Tt +1 – 0.00 0.45 0.69 0.07 0.07 1.97 0.95 1.05 2.28 1.76 1.77 1.36 ADJUSTED FORECAST AFt +1 – 37.00 38.95 40.44 38.44 38.44 47.82 45.37 46.76 58.13 53.19 54.98 54.96

T3

= (F3 - F2) + (1 - ) T2 = (0.30)(38.5 - 37.0) + (0.70)(0) = 0.45

AF3 = F3 + T3 = 38.5 + 0.45 = 38.95 T13 = (F13 - F12) + (1 - ) T12 = (0.30)(53.61 - 53.21) + (0.70)(1.77) = 1.36 AF13 = F13 + T13 = 53.61 + 1.36 = 54.96

Adjusted Exponential Smoothing Forecasts
70 – Adjusted forecast ( 60 – Actual 50 – 40 – 30 – 20 – 10 – 0– | 1 | 2 | 3 | 4 | 5 | | 6 7 Period | 8 | 9 | 10 | 11 | 12 | 13 Forecast ( = 0.50) =

Linear Trend Line
xy - nxy x2 - nx2 a = y-bx b = where n = number of periods x x = = mean of the x values n y y = n = mean of the y values

y = a + bx
where a = intercept b = slope of the line x = time period y = forecast for demand for period x

Demand

Least Squares Example
x(PERIOD) 1 2 3 4 5 6 7 8 9 10 11 12 78 y(DEMAND) 73 40 41 37 45 50 43 47 56 52 55 54 557 xy 37 80 123 148 225 300 301 376 504 520 605 648 3867 x2 1 4 9 16 25 36 49 64 81 100 121 144 650

Least Squares Example (cont.)

78 = 6.5 12 557 y = = 46.42 12 xy - nxy b = = x2 - nx2 x =

3867 - (12)(6.5)(46.42) =1.72 650 - 12(6.5)2

a = y - bx = 46.42 - (1.72)(6.5) = 35.2

Linear trend line y = 35.2 + 1.72x Forecast for period 13 y = 35.2 + 1.72(13) = 57.56 units
70 – 60 – 50 – Actual

Seasonal Adjustments
The multiplicative seasonal model can modify trend data to accommodate seasonal variations in demand
1. Find average historical demand for each season

Demand

40 – Linear trend line 30 – 20 – 10 – 0– | 1 | 2 | 3 | 4 | 5 | | 6 7 Period | 8 | 9 | 10 | 11 | 12 | 13

2. Compute the average demand over all seasons 3. Compute a seasonal index for each season 4. Estimate next year’s total demand year’ 5. Divide this estimate of total demand by the number of seasons, then multiply it by the seasonal index for that season

Seasonal Adjustments

Seasonal Adjustment (cont.)
YEAR DEMAND (1000’S PER QUARTER) (1000’ 1 2 3 4 Total 12.6 14.1 15.3 42.0 8.6 10.3 10.6 29.5 6.3 7.5 8.1 21.9 17.5 18.2 19.6 55.3 45.0 50.1 53.6 148.7

Repetitive increase/ decrease in demand Use seasonal factor to adjust forecast Di D

2002 2003 2004 Total

Seasonal factor = Si =

S1 = S2 =

D1 42.0 = = 0.28 D 148.7 D2 29.5 = = 0.20 D 148.7

S3 = S4 =

D3 21.9 = = 0.15 D 148.7 D4 55.3 = = 0.37 D 148.7

Seasonal Adjustment (cont.)
For 2005 y = 40.97 + 4.30x = 40.97 + 4.30(4) = 58.17 4.30x

Forecast Accuracy
Forecast error
difference between forecast and actual demand MAD
mean absolute deviation

SF1 = (S1) (F5) = (0.28)(58.17) = 16.28 (S (F SF2 = (S2) (F5) = (0.20)(58.17) = 11.63 (S (F SF3 = (S3) (F5) = (0.15)(58.17) = 8.73 (S (F SF4 = (S4) (F5) = (0.37)(58.17) = 21.53 (S (F

MAPD
mean absolute percent deviation

Cumulative error Average error or bias

Mean Absolute Deviation (MAD)
PERIOD 1 2 3 4 5 6 7 8 9 10 11 12 37 40 41 37 MAD 45 50 43 47 56 52 55 54 557

MAD Example
DEMAND, Dt Ft ( =0.3) (Dt - Ft) – 3.00 3.10 -1.83 6.72 9.69 -0.20 3.86 11.70 4.19 5.94 3.15 49.31 |Dt - Ft| – 3.00 3.10 1.83 6.72 9.69 0.20 3.86 11.70 4.19 5.94 3.15 53.39 37.00 37.00 37.90 D38.83 t t - F 38.28 n 40.29 53.39 43.20 43.14 11 44.30 4.85 47.81 49.06 50.84

MAD =
where

Dt - Ft n

= = =

t = period number Dt = demand in period t Ft = forecast for period t n = total number of periods = absolute value

Other Accuracy Measures
Mean absolute percent deviation (MAPD) MAPD = Cumulative error E = et Average error et E= n |Dt - Ft| Dt

Comparison of Forecasts

FORECAST Exponential smoothing ( = 0.30) Exponential smoothing ( = 0.50) Adjusted exponential smoothing ( = 0.50, = 0.30) Linear trend line

MAD 4.85 4.04 3.81 2.29

MAPD 9.6% 8.5% 7.5% 4.9%

E 49.31 33.21 21.14 –

(E) 4.48 3.02 1.92 –

Forecast Control
Tracking signal
monitors the forecast to see if it is biased high or low Tracking signal = (Dt - Ft) E = MAD MAD
PERIOD

Tracking Signal Values
DEMAND Dt FORECAST, Ft ERROR Dt - Ft E= (Dt - Ft) MAD TRACKING SIGNAL

1 MAD 0.8 Control limits of 2 to 5 MADs are used most frequently

1 2 3 4 5 6 7 8 9 10 11 12

37 40 41 37 45 50 43 47 56 52 55 54

37.00 – – 37.00 3.00 3.00 37.90 3.10 6.10 38.83 -1.83 4.27 38.28 6.72 10.99 Tracking signal for period 3 40.29 9.69 20.68 43.20 -0.20 6.10 20.48 43.14 TS3 = 3.86 =24.34 2.00 3.05 36.04 44.30 11.70 47.81 4.19 40.23 49.06 5.94 46.17 50.84 3.15 49.32

– 3.00 3.05 2.64 3.66 4.87 4.09 4.06 5.01 4.92 5.02 4.85

– 1.00 2.00 1.62 3.00 4.25 5.01 6.00 7.19 8.18 9.20 10.17

Tracking Signal Plot
3 – Tracking signal (MAD) 2 – 1 – 0 – -1 – -2 – -3 – | 0 | 1 | 2 | 3 | 4 | 5 | 6 Period | 7 | 8 | 9 | 10 | 11 | 12 Linear trend line Exponential smoothing ( = 0.30)

Statistical Control Charts
(Dt - Ft)2 n-1

=

Using we can calculate statistical control limits for the forecast error Control limits are typically set at 3

Statistical Control Charts
18.39 – UCL = +3 12.24 – 6.12 – Errors 0– -6.12 – -12.24 – -18.39 – | 0 LCL = -3 | 1 | 2 | 3 | 4 | 5 | 6 Period | 7 | 8 | 9 | 10 | 11 | 12

Regression Methods
Linear regression
a mathematical technique that relates a dependent variable to an independent variable in the form of a linear equation

Correlation
a measure of the strength of the relationship between independent and dependent variables

Linear Regression
y = a + bx
where a = intercept b = slope of the line x = y = x = mean of the x data n y n = mean of the y data a = y-bx xy - nxy b = x2 - nx2

Linear Regression Example
x (WINS) 4 6 6 8 6 7 5 7 49 y (ATTENDANCE) 36.3 40.1 41.2 53.0 44.0 45.6 39.0 47.5 346.7 xy 145.2 240.6 247.2 424.0 264.0 319.2 195.0 332.5 2167.7 x2 16 36 36 64 36 49 25 49 311

Linear Regression Example (cont.)
49 = 6.125 8 346.9 y= = 43.36 8 x= b=

Linear Regression Example (cont.)
Regression equation y = 18.46 + 4.06x
60,000 – 50,000 – 40,000 – 30,000 – 20,000 – 10,000 – | 0 | 1 | 2 | 3 | 4 | | 5 6 Wins, x | 7 | 8 | 9 | 10

Attendance forecast for 7 wins y = 18.46 + 4.06(7) = 46.88, or 46,880

=

(2,167.7) - (8)(6.125)(43.36) (311) - (8)(6.125)2

Attendance, y

xy - nxy2 x2 - nx2

Linear regression line, y = 18.46 + 4.06x 4.06x

= 4.06 a = y - bx = 43.36 - (4.06)(6.125) = 18.46

Correlation and Coefficient of Determination
Correlation, r Correlation,
Measure of strength of relationship Varies between -1.00 and +1.00

Computing Correlation
r= n xy [n x2 -( x)2] x y [n y2 - ( y)2] [n

Coefficient of determination, determination,

r2

r=

(8)(2,167.7) - (49)(346.9) [(8)(311) - (49)2] [(8)(15,224.7) - (346.9)2] r = 0.947 Coefficient of determination r2 = (0.947)2 = 0.897

Percentage of variation in dependent variable resulting from changes in the independent variable

Multiple Regression
Study the relationship of demand to two or more independent variables
y= 0+ where
1, 1x1 + 2x2 …

+

kx k

THE END

= the intercept = parameters for the independent variables x1, … , xk = independent variables
0 k

…,

Exponential Smoothing
• EXERCISE : TUTORIAL
Form of weighted moving average
Weights decline exponentially Most recent data weighted most

Requires smoothing constant ( )
Ranges from 0 to 1 Subjectively chosen

Involves little record keeping of past data

Exponential Smoothing
New forecast = last period’s forecast period’ + (last period’s actual demand period’ – last period’s forecast) period’ forecast) Ft = Ft – 1 + (At – 1 - Ft – 1)
where Ft = new forecast Ft – 1 = previous forecast = smoothing (or weighting) constant (0 1)

Exponential Smoothing
Example
Predicted demand = 142 Ford Mustangs Actual demand = 153 Smoothing constant = .20

Exponential Smoothing
Example
Predicted demand = 142 Ford Mustangs Actual demand = 153 Smoothing constant = .20 New forecast = 142 + .2(153 – 142)

Exponential Smoothing
Example
Predicted demand = 142 Ford Mustangs Actual demand = 153 Smoothing constant = .20 New forecast = 142 + .2(153 – 142) = 142 + 2.2 = 144.2 144 cars

Effect of Smoothing Constants
225 – Weight Assigned to Demand
Most Recent Period ( ) .1 .5 2nd Most 3rd Most 4th Most 5th Most Recent Recent Recent Recent Period Period Period Period 2 3 (1 - ) (1 - ) (1 - ) (1 - )4 .09 .25 .081 .125 .073 .063 .066 .031

Impact of Different

200 –

Actual demand

= .5

Smoothing Constant = .1 = .5

175 –

= .1
150 – | 1 | 2 | 3 | 4 | 5 Quarter | 6 | 7 | 8 | 9

Choosing
The objective is to obtain the most accurate forecast no matter the technique
We generally do this by selecting the model that gives us the lowest forecast error
Forecast error = Actual demand - Forecast value = At - Ft

Common Measures of Error
Mean Absolute Deviation (MAD) MAD)
MAD = |actual - forecast| n

Mean Squared Error (MSE) MSE)
MSE = (forecast errors)2 errors) n

Common Measures of Error
Mean Absolute Percent Error (MAPE) MAPE)
n

Comparison of Forecast Error
Quarter Actual Tonnage Unloaded Rounded Forecast with = .10 Absolute Deviation for = .10 Rounded Forecast with = .50 Absolute Deviation for = .50

100 MAPE =

i=1

|actuali - forecasti|/actuali n

1 2 3 4 5 6 7 8

180 168 159 175 190 205 180 182

175 176 175 173 173 175 178 178

5 8 16 2 17 30 2 4 84

175 178 173 166 170 180 193 186

5 10 14 9 20 25 13 4 100

Comparison of Forecast Error
MAD = Actual
Quarter Tonage Unloaded

Comparison of Forecast Error
(forecast errors)2 Rounded Absolute MSE = Actual Forecast Deviation n Tonage with for
Quarter Unloaded = .10 = .10 Rounded Forecast with = .50 Absolute Deviation for = .50

|deviations| Rounded Absolute Forecast Deviation n with for
= .10 = .10

1 2 3 4 For 5 6 7 8

For

180 175 5 168 = 84/8 = 10.50 8 176 159 175 16 175.50 173 2 = 190 173 17 205 = 100/8 = 12.50 175 30 180 178 2 182 178 4 84

= .10

Rounded Forecast with = .50

Absolute Deviation for = .50

175 178 173 166 170 180 193 186

5 10 14 9 20 25 13 4 100

1 2 3 4 For 5 6 7 8

For

180 175 168 1,558/8 = 176 = 159 175 175.50 173 = 190 173 = 2051,612/8 = 175 180 178 182 178

= .10

5 8 194.75

16 2 17 201.50 30 2 4 84 MAD 10.50

175 178 173 166 170 180 193 186

5 10 14 9 20 25 13 4 100 12.50

Comparison of Forecast Error n
100 |deviationi|/actuali Absolute Rounded i =Rounded 1 MAPE = Actual Forecast Deviation Forecast Tonage with n for with Quarter Unloaded = .10 = .10 = .50 For 180 .10 175 = 1 5 175 2 168 176 8 178 = 45.62/8 = 5.70%
3 4 5 6 7 8 159 175 16 2 For 175 .50 173 = 190 173 17 = 54.8/8 = 6.85% 205 175 30 180 182 178 178 MAD MSE 2 4 84 10.50 194.75 173 166 170 180 193 186
Absolute Deviation for = .50

Comparison of Forecast Error
Quarter Actual Tonnage Unloaded Rounded Forecast with = .10 Absolute Deviation for = .10 Rounded Forecast with = .50 Absolute Deviation for = .50

5 10 14 9 20 25 13 4 100 12.50 201.50

1 2 3 4 5 6 7 8

180 168 159 175 190 205 180 182

175 176 175 173 173 175 178 178 MAD MSE MAPE

5 8 16 2 17 30 2 4 84 10.50 194.75 5.70%

175 178 173 166 170 180 193 186

5 10 14 9 20 25 13 4 100 12.50 201.50 6.85%

Exponential Smoothing with Trend Adjustment
When a trend is present, exponential smoothing must be modified
exponentially exponentially Forecast including (FITt) = smoothed (Ft) + (Tt) smoothed trend forecast trend

Exponential Smoothing with Trend Adjustment
Ft = (At - 1) + (1 - )(Ft - 1 + Tt - 1) )(F Tt = (Ft - Ft - 1) + (1 - )Tt - 1
Step 1: Compute Ft Step 2: Compute Tt Step 3: Calculate the forecast FITt = Ft + Tt

Exponential Smoothing with Trend Adjustment Example
Month(t) Month( 1 2 3 4 5 6 7 8 9 10
Table 4.1

Exponential Smoothing with Trend Adjustment Example
Month(t) Month( 1 2 3 4 5 6 7 8 9 10
Table 4.1

Actual Demand (At) 12 17 20 19 24 21 31 28 36

Smoothed Forecast, Ft 11

Smoothed Trend, Tt 2

Forecast Including Trend, FITt 13.00

Forecast Actual Smoothed Smoothed Including Demand (At) Forecast, Ft Trend, Tt Trend, FITt 12 11 2 13.00 17 20 19 Step 1: Forecast for Month 2 24 21 F2 = A1 + (1 - )(F1 + T1) 31 28 F2 = (.2)(12) + (1 - .2)(11 + 2) 36

= 2.4 + 10.4 = 12.8 units

Exponential Smoothing with Trend Adjustment Example
Month(t) Month( 1 2 3 4 5 6 7 8 9 10
Table 4.1

Exponential Smoothing with Trend Adjustment Example
Month(t) Month( 1 2 3 4 5 6 7 8 9 10
Table 4.1

Forecast Actual Smoothed Smoothed Including Demand (At) Forecast, Ft Trend, Tt Trend, FITt 12 11 2 13.00 17 12.80 20 19 Step 2: Trend for Month 2 24 21 T2 = (F2 - F1) + (1 - )T1 31 28 T2 = (.4)(12.8 - 11) + (1 - .4)(2) 36

= .72 + 1.2 = 1.92 units

Forecast Actual Smoothed Smoothed Including Demand (At) Forecast, Ft Trend, Tt Trend, FITt 12 11 2 13.00 17 12.80 1.92 20 19 Step 3: Calculate FIT for Month 2 24 21 FIT2 = F2 + T1 31 28 FIT2 = 12.8 + 1.92 36

= 14.72 units

Exponential Smoothing with Trend Adjustment Example
Month(t) Month( 1 2 3 4 5 6 7 8 9 10
Table 4.1

Exponential Smoothing with Trend Adjustment Example
35 – 30 – Product demand 25 – 20 – 15 – 10 – 5 – 0 – | 1 | 2 | 3 | | | 4 5 6 Time (month) | 7 | 8 | 9

Actual Demand (At) 12 17 20 19 24 21 31 28 36

Smoothed Forecast, Ft 11 12.80 15.18 17.82 19.91 22.51 24.11 27.14 29.28 32.48

Smoothed Trend, Tt 2 1.92 2.10 2.32 2.23 2.38 2.07 2.45 2.32 2.68

Forecast Including Trend, FITt 13.00 14.72 17.28 20.14 22.14 24.89 26.18 29.59 31.60 35.16

Actual demand (At)

Forecast including trend (FITt)

Figure 4.3

Trend Projections
Fitting a trend line to historical data points to project into the medium-to-long-range medium- to- longLinear trends can be found using the least squares technique
^ y = a + bx
^ where y = computed value of the variable to be predicted (dependent variable) a = y-axis intercept yb = slope of the regression line x = the independent variable Values of Dependent Variable

Least Squares Method
Actual observation (y value)
Deviation5 Deviation3 Deviation4 Deviation1 Deviation2 Deviation7

Deviation6

^ Trend line, y = a + bx

Time period

Figure 4.4

Least Squares Method
Values of Dependent Variable Actual observation (y value)
Deviation5 Deviation3 Deviation7

Least Squares Method
Equations to calculate the regression variables

Deviation6

^ y = a + bx

Least squares method minimizes the sum of the Deviation squared errors (deviations)
4

b=

Deviation1 Deviation2

xy - nxy x2 - nx2

^ Trend line, y = a + bx

a = y - bx
Time period
Figure 4.4

Least Squares Example
Year 1999 2000 2001 2002 2003 2004 2005 Time Period (x) 1 2 3 4 5 6 7 x = 28 x=4 b= Electrical Power Demand 74 79 80 90 105 142 122 y = 692 y = 98.86 x2 1 4 9 16 25 36 49 x2 = 140 xy 74 158 240 360 525 852 854 xy = 3,063 Year

Least Squares Example
Time Period (x) Electrical Power Demand x2 1 4 9 16 25 36 49 x2 = 140 xy 74 158 240 360 525 852 854 xy = 3,063 1999 1 74 2000 2 79 The trend line is 80 2001 3 2002 4 90 ^= 2003 105 y 5 56.70 + 10.54x 2004 6 142 2005 7 122 x = 28 y = 692 x=4 y = 98.86 b=

3,063 - (7)(4)(98.86) xy - nxy = = 10.54 10.54 140 - (7)(42) x2 - nx2

3,063 - (7)(4)(98.86) xy - nxy = = 10.54 10.54 140 - (7)(42) x2 - nx2

a = y - bx = 98.86 - 10.54(4) = 56.70

a = y - bx = 98.86 - 10.54(4) = 56.70

Least Squares Example
160 150 140 130 120 110 100 90 80 70 60 50 – – – – – – – – – – – – | 1999 | 2000 | 2001

Least Squares Requirements
1. We always plot the data to insure a linear relationship 2. We do not predict time periods far beyond the database 3. Deviations around the least squares line are assumed to be random

Trend line, ^ y = 56.70 + 10.54x

Power demand

| 2002

| 2003 Year

| 2004

| 2005

| 2006

| 2007

Seasonal Variations In Data
The multiplicative seasonal model can modify trend data to accommodate seasonal variations in demand
1. Find average historical demand for each season 2. Compute the average demand over all seasons 3. Compute a seasonal index for each season 4. Estimate next year’s total demand year’ 5. Divide this estimate of total demand by the number of seasons, then multiply it by the seasonal index for that season
Month Jan Feb Mar Apr May Jun Jul Aug Sept Oct Nov Dec

Seasonal Index Example
Demand 2003 2004 2005 80 70 80 90 113 110 100 88 85 77 75 82 85 85 93 95 125 115 102 102 90 78 72 78 105 85 82 115 131 120 113 110 95 85 83 80 Average 2003-2005 200390 80 85 100 123 115 105 100 90 80 80 80 Average Monthly 94 94 94 94 94 94 94 94 94 94 94 94 Seasonal Index

Seasonal Index Example
Month Demand 2003 2004 2005 Average 2003-2005 2003Average Monthly Seasonal Index 0.957 Month Jan Feb Mar Apr May Jun Jul Aug Sept Oct Nov Dec Jan 80 85 105 90 94 Feb 70 85 85 80 94 Mar 80 93 average 2003-2005 monthly demand 82 85 94 Seasonal index = 115 Apr 90 95 100 94 average monthly demand May 113 125 131 123 94 = 90/94 = .957 Jun 110 115 120 115 94 Jul 100 102 113 105 94 Aug 88 102 110 100 94 Sept 85 90 95 90 94 Oct 77 78 85 80 94 Nov 75 72 83 80 94 Dec 82 78 80 80 94

Seasonal Index Example
Demand 2003 2004 2005 80 70 80 90 113 110 100 88 85 77 75 82 85 85 93 95 125 115 102 102 90 78 72 78 105 85 82 115 131 120 113 110 95 85 83 80 Average 2003-2005 200390 80 85 100 123 115 105 100 90 80 80 80 Average Monthly 94 94 94 94 94 94 94 94 94 94 94 94 Seasonal Index 0.957 0.851 0.904 1.064 1.309 1.223 1.117 1.064 0.957 0.851 0.851 0.851

Seasonal Index Example
Month Jan Feb Mar Apr May Jun Jul Aug Sept Oct Nov Dec Demand 2003 2004 2005 Average 2003-2005 2003Average Monthly Seasonal Index 0.957 0.851 0.904 1.064 1.309 1.223 1.117 1.064 0.957 0.851 0.851 0.851 80 85 105 90 94 70 85 Forecast for 2006 85 80 94 80 93 82 85 94 Expected annual demand = 1,200 90 95 115 100 94 113 125 131 123 94 110 115 120 1,200 115 94 Jan x .957 = 96 94 100 102 113 12 105 88 102 110 100 94 1,200 85 90 Feb 95 x90 .851 = 85 94 77 78 85 12 80 94 75 72 83 80 94 82 78 80 80 94

Seasonal Index Example
140 – 130 – Demand 120 – 110 – 100 – 90 – 80 – 70 – | J | F | M | A | M | | J J Time | A | S | O | N | D

2006 Forecast 2005 Demand 2004 Demand 2003 Demand

San Diego Hospital
Trend Data
Index for Inpatient Days 10,200 – Inpatient Days 10,000 – 9,800 – 9,600 – 9530 9,400 – 9,200 – | | | | | | | | | | | 9,000 – | Jan Feb Mar Apr May June July Aug Sept Oct Nov Dec 67 68 69 70 71 72 73 74 75 76 77 78 Month
Figure 4.6

San Diego Hospital
Seasonal Indices
1.06 – 1.04 – 1.02 – 1.00 – 0.98 – 0.96 – 0.94 – 0.97 1.04 1.02 1.01 0.99 0.99 0.97 0.96 1.00 0.98 1.03 1.04

9573 9551 9594

9616 9637

9659 9680

9702 9723

9745 9766

| | | | | | | | | | | 0.92 – | Jan Feb Mar Apr May June July Aug Sept Oct Nov Dec 67 68 69 70 71 72 73 74 75 76 77 78 Month
Figure 4.7

San Diego Hospital
Combined Trend and Seasonal Forecast
10,200 – Inpatient Days 10,000 – 9911 9,800 – 9,600 – 9,400 – 9,200 – 9,000 – 9265 9520 9542 9411 9355 9764 9691 9949 9724 9572 10068

Associative Forecasting
Used when changes in one or more independent variables can be used to predict the changes in the dependent variable Most common technique is linear regression analysis We apply this technique just as we did in the time series example

| | | | | | | | | | | | Jan Feb Mar Apr May June July Aug Sept Oct Nov Dec 67 68 69 70 71 72 73 74 75 76 77 78 Month
Figure 4.8

Associative Forecasting
Forecasting an outcome based on predictor variables using the least squares technique
^ y = a + bx
^ where y = computed value of the variable to be predicted (dependent variable) a = y-axis intercept yb = slope of the regression line x = the independent variable though to predict the value of the dependent variable

Associative Forecasting
Example
Sales ($000,000), y 2.0 3.0 2.5 2.0 2.0 3.5 Local Payroll ($000,000,000), x 1 3 4 4.0 – 2 1 3.0 – 7
Sales 2.0 – 1.0 – 0 | 1 | 2 | | | | 3 4 5 6 Area payroll | 7

Associative Forecasting
Example
Sales, y 2.0 3.0 2.5 2.0 2.0 3.5 y = 15.0 Payroll, x 1 3 4 2 1 7 x = 18 x2 1 9 16 4 1 49 x2 = 80 xy 2.0 9.0 10.0 4.0 2.0 24.5 xy = 51.5

Associative Forecasting
Example
^ y = 1.75 + .25x .25x If payroll next year is estimated to be $600 million, then: Sales = 1.75 + .25(6) Sales = $325,000

Sales = 1.75 + .25(payroll) .25(payroll)
4.0 – Sales 3.25 3.0 – 2.0 – 1.0 – 0 | 1 | 2 | | | | 3 4 5 6 Area payroll | 7

x = x/6 = 18/6 = 3 y = y/6 = 15/6 = 2.5

b=

51.5 - (6)(3)(2.5) xy - nxy = = .25 80 - (6)(32) x2 - nx2

a = y - bx = 2.5 - (.25)(3) = 1.75

Standard Error of the Estimate
A forecast is just a point estimate of a future value This point is actually the mean of a probability distribution

Standard Error of the Estimate
(y - yc)2 n-2

Sy,x =
4.0 – Sales 3.25 3.0 – 2.0 – 1.0 – 0 | 1 | 2 | | | | 3 4 5 6 Area payroll | 7

where y = y-value of each data point yc = computed value of the dependent variable, from the regression equation n = number of data points

Figure 4.9

Standard Error of the Estimate
Computationally, this equation is considerably easier to use Sy,x = y2 - a y - b xy n-2

Standard Error of the Estimate
Sy,x = y2 - a y - b xy = n-2
4.0 – Sales 3.25 3.0 – 2.0 – 1.0 – 0 | 1 | 2 | | | | 3 4 5 6 Area payroll | 7

39.5 - 1.75(15) - .25(51.5) 6-2

Sy,x = .306 The standard error of the estimate is $30,600 in sales

We use the standard error to set up prediction intervals around the point estimate

Correlation
How strong is the linear relationship between the variables? Correlation does not necessarily imply causality! Coefficient of correlation, r, measures degree of association
Values range from -1 to +1

Correlation Coefficient
r= n xy - x y [n x2 - ( x)2][n y2 - ( y)2] ][n

y

y Correlation Coefficient

Correlation
Coefficient of Determination, r2, measures the percent of change in y predicted by the change in x
Values range from 0 to 1 Easy to interpret

r= [n x2 (a) Perfect positive x correlation: r = +1
y

n xy - x y - ( x)2][n y2 Positive)2] ][n (b) - ( y
correlation: 0<r<1

x

y

For the Nodel Construction example: r = .901
(c) No correlation: r=0

x

(d) Perfect negative x correlation: r = -1

r2 = .81

Multiple Regression Analysis
If more than one independent variable is to be used in the model, linear regression can be extended to multiple regression to accommodate several independent variables ^ y = a + b1x1 + b2x2 … Computationally, this is quite complex and generally done on the computer

Multiple Regression Analysis
In the Nodel example, including interest rates in the model gives the new equation: ^ y = 1.80 + .30x1 - 5.0x2 .30x 5.0x An improved correlation coefficient of r = .96 means this model does a better job of predicting the change in construction sales Sales = 1.80 + .30(6) - 5.0(.12) = 3.00 Sales = $300,000

Monitoring and Controlling Forecasts
Tracking Signal
Measures how well the forecast is predicting actual values Ratio of running sum of forecast errors (RSFE) to mean absolute deviation (MAD)
Good tracking signal has low values If forecasts are continually high or low, the forecast has a bias error

Monitoring and Controlling Forecasts
RSFE Tracking = signal MAD (actual demand in period i forecast demand in period i) |actual - forecast|/n) forecast|/n)

Tracking = signal

Tracking Signal
Signal exceeding limit Tracking signal + 0 MADs – Upper control limit

Tracking Signal Example
Actual Qtr Demand Forecast Demand Error RSFE Absolute Forecast Error Cumulative Absolute Forecast Error MAD

Acceptable range

1 2 3 4 5 6

90 95 115 100 125 140

100 100 100 110 110 110

-10 -5 +15 -10 +15 +30

-10 -15 0 -10 +5 +35

10 5 15 10 15 30

10 15 30 40 55 85

10.0 7.5 10.0 10.0 11.0 14.2

Lower control limit Time

Tracking Signal Example
Tracking
Actual Signal Forecast (RSFE/MAD) Qtr Demand Demand Error RSFE Absolute Forecast Error Cumulative Absolute Forecast Error MAD

Adaptive Forecasting
It’s possible to use the computer to It’ continually monitor forecast error and adjust the values of the and coefficients used in exponential smoothing to continually minimize forecast error This technique is called adaptive smoothing

1 2 3 4 5 6

90 100 -1 -10 -10/10 = 95 100 -2 -5 -15/7.5 = 115 0/10 = 0 +15 100 100 110 -1 -10 -10/10 = 125 110 +15 +5/11 = +0.5 140 110 +2.5 +35/14.2 = +30

-10 -15 0 -10 +5 +35

10 5 15 10 15 30

10 15 30 40 55 85

10.0 7.5 10.0 10.0 11.0 14.2

The variation of the tracking signal between -2.0 and +2.5 is within acceptable limits

Focus Forecasting
Developed at American Hardware Supply, focus forecasting is based on two principles:
1. Sophisticated forecasting models are not always better than simple models 2. There is no single techniques that should be used for all products or services This approach uses historical data to test multiple forecasting models for individual items The forecasting model with the lowest error is then used to forecast the next demand

Sign up to vote on this title
UsefulNot useful