Dispersion σ Standard Forecast - A prediction of future events used for planning purposes

Actual - of Forecast % |Error| Deviation of Time series -The repeated observations of demand for a service or
Forecast Error /Actual Errors product in their order of occurrence Demand Pattern
Horizontal, Trend, Seasonal, Cyclical, Random
Blank Actual Forecast Error E |Error| Error 2 % Error Error - (Error - )2 Type of Forecasting Technique -Judgment methods, Causal
Period 1 39 41 −2 2 4 5.13% methods, Time-series analysis’ Trend projection using regression
Period n 37 43 −6 6 36 16.22%
Average Σ Actual / n Forecast Error Et = Dt - Ft (D Actual demand and F Forecasted
CFC Cumulative Forecast Error or Bias Error Σ Error Demand in period t)
Σ Error / n or The MSE, σ, MAD statistics provide measures of forecast error
AFE Avg Forecast Error or Mean Bias
CFE / n variability. IF MSE, MAD or σ small->forecast close to actual
MAD Mean Absolute Deviation (Error) Σ |Error| / n demand
MSE Mean Squared Error Σ Error 2 / CFE<0 => Forecast Over (Et = Dt - Ft)
Σ % Error / CFE>0 => Forecast Under (Et = Dt - Ft)
MAPE Mean Absolute Percent Error MAPE-forecast error was % of actual demand.
n
These measures become more reliable as the number of periods of
 
2
 Et  E data increases
σ Standard Deviation of Errors  
n 1
Judgmental forecasts use contextual knowledge gained through
SE Standard Error experience. Salesforce estimates, Executive opinion, Market
research, Delphi method
Dependent variable – The variable that one wants to forecast
Independent variable – The variable that is assumed to affect
the dependent variable and thereby “cause” the results observed
in the past

∆y y2-y1
Slope = or
∆x x2-x1

Average
Seasonal Seasonal Year 3
Seasonal
Quarter Year 1 Factor Year 2 Factor Expected Year 3 Forecast
Factor
SF1 SF2 Sales
Avg SF
1 45 45/Avg 67 67/280.75 (SF1+SF2) / 2 Avg SF* Year3 Avg
Total 45 67 1,850
Naïve forecast -The forecast for the next period equals the demand for the
Average 11.25 16.75 462.50
current period (Forecast = Dt)
Ft=Dt-1
Simple moving avg, Weighted moving avg, Exponential smoothing Exponential Smoothing weighted moving Multiplicative seasonal method - Calculate for each
average that calculates the average of a time Annual Demand of Sum of all demands for all
series by implicitly giving recent demands more an year= seasons for that year
weight than earlier demands Average demand for Annual Demand
Note: If F1 , Ft-1 ….Ft is not given then use F1 =D1 each season= # of seasons/year
Actual Demand for the
Seasonal Factor
season
(Index) for a season= Avg demand for the season
Σ Seasonal Factors for the
Avg Seasonal Factor
Ft 1  W1D1  W2D2      Wn Dt  n 1 season over the year
(Index) for a season=
Highest weight goes to most recent data; sum of the weights=1 # of years (for the season)
Avg Forecast for Yearly forecast
Linear trent regression equation is Yt = a + bt each season= # of seasons per year
Replace x by time period t in SLR model Avg Forecast for each season
Seasonal Forecast
* Avg Seasonal Factor (Index)
Dynamic demand--use higher α or smaller n

for a season=
for a season
Stable demand-use lower α or larger n

Forecast Error Ft ~ N (0, σ2)

σ 1.25 MAD or MAD 0.8σ

-1.5≤TS≤1.5 is the desired TS range, else model needs to be changed

 Process Alpha Process Beta Week 1
Total Value (Output) = Units Produced * Value Per Unit A B Total A B Total Units Produced 1,124
Units Produced 50 60 30 80 Labor ($) 12,735
Total Value (Input) = Labor + Materials + Overhead Materials ($) 21,041
Value Per Unit $175 $140 $175 $140
Productivity = Output Overhead ($) 8,992
Total Value (Output) 8,750 8,400 17,150 5,250 11,200 16,450 Value Per Unit $100
Input
Labor ($) 1,200 1,400 2,600 1,000 2,000 3,000 Total Value (Output) (1) 112,400
Improvement = Last Period - First Period * 100 Materials ($) 2,500 3,000 5,500 1,400 3,500 4,900 Total Value (Input) (2) 42,768
First Period Overhead ($) 6,000 6,000 5,000 5,000 Multifactor Productivity (3)
2.628
Productivity of a process = Output (Total value of output A & B) Total Value (Input) 14,100 12,900 Cost of Labor Per Unit $50
Input (Total Input of A & B) Multifactor Productivity 1.216 1.275 Labor (hrs) 254.70
4.41
Labor Productivity (Units/Hr) 6.596 5.483 Labor Productivity (Units/Hr)
If annual Decision theory Row
Total Cost = Fixed Cost + Variable Cost
demand ≤ Average/
Total Cost = F + cQ Alternative Low Moderate High Min Max
where Variable Cost is Variable Cost per Unit BEQ, then Weighted
buy,
c times Quantity Q Hire ($250,000) $100,000
$625,000 ($250,000) $625,000 $158,333
otherwise
Total Revenue = Price * Quantity Subcontract $100,000 $150,000
$415,000 $100,000 $415,000 $221,667
make (use
Total Revenue = pQ Do Nothing $50,000 $80,000
$300,000 $50,000 $300,000 $143,333
process
At Break Even, Total Revenue = Total Cost
with higher Maximin $100,000
pQ = F + cQ
variable (or Maximax $625,000
Break Even Quantity lower fixed LaPlace $221,667
cost) Maximin - Minimum payoff for each alternative, then choose alternative with maximum
Maximax - Maximum of each alternative and then select maximum amongst
Preference Matrix
Best decision : Hire ($625,000)
Laplace - Average of all the alternatives and then select the best payoff
Factor
Performance Score Score Weighted Weighted
Weight Score Score Best decision : Subcontract ($221,667)
Criteria A B
W Ax W BxW
Minimax Regret - Find regrets for each alternative, selcting alternative that helps in
Total 100
minimizing maximum regret (difference between the highest value and each alternative)

Decision Tree – Opportunity Loss Row

Decisions made Alternative Low Moderate High Max
Optimistic outlook- MaxiMax
sequentially Hire $350,000 $50,000 $ - $350,000 Pessimistic outlook-MaxiMin
Subcontract $ - $ - $210,000 $210,000
Do Nothing $50,000 $70,000 $325,000 $325,000
Minimax $210,000

Core Competencies -Technical knowhow, Reliable internal process, Closed external relationships

Hurwicz criteria: α: coefficient of optimism. 1-α: coeff of pessimism

Maximin (best of worst) Maximax (Best of Best). Laplace (realist or best avg outcome) Minimax regret (best of the worst regret)