Time Series and Forecasng – uncover

and extrapolate the paern in the

me series to forecast the future

Trend (T) - long terms paerns

Slope - y is increasing over me > 0

y is decreasing over me < 0

Seasonality (S) – systemac paern

that keeps happening every year

Random (R) – beyond our control;

uctuaons, outliers

Cycle (C) – economic cycle; one cycle

is recession, recover, inaon, recover

Qualitave – involve the use of expect

judgement; data unavailable, less than

two years of data, no paern, has

signi"cant level changes in the data

Conrm – top management, sales

people, customers, experts, delphi

Delphi Model - short meeng with

key people to contribute and

con"rm analysis; need consensus

Quantave – informaon can be

quan"ed, and it is reasonable to

assume the paern of the past will

connue into the future; minimum

two years of data, paern in the data,

no signi"cant level changes in the data

Trend – the long run shi$ or

movement in the me series

observable over several periods of

me

Seasonal Pa#ern – component of me

series that shows a periodic paern

over one year or less

Cyclical Pa#ern – results in period

above and below trend behavior of

the me series last more than one

year

Forecast Error – the amount by which

the forecasted value % di&ers from the

observed value for me period t

e = y

- %

Posive Error – forecasng method

underesmated the actual value

Negave Error – forecasng method

overesmated the actual value

Mean Absolute Deviaon (MAD) –

measure of forecasng accuracy

∑∨e∨

Lowest measure of accuracy for the

best model; small MAD

Mean Squared Error (MSE) – average

of the sum of the squared di&erences

between the forecast values and the

actual me series values

∑

Naїve Model – uses the value of the

me series from the most recent

period as the forecast for the current

period

SMA

– Simple Moving Average with

one period in Excel

Moving Average – every me a new

observaon becomes available for the

me series, it replaces the oldest

observaon in the equaon and a new

average is computer

SMA – Simple Moving Average

WMA – Weighted Moving Average;

Enter weights into excel from largest

To smallest

Exponenal Smoothing – uses a

weighted average of past me series

values as the forecast

t+1

= αy

+ (1 – α)%

Change inial forecast value if given

Smoothing Constant – provides the

weight given to the most recent me

series values in the calculaon of the

forecast value

0 < α < 1

Linear Trend – mathemacal equaon

showing how variables are related

Autoregressive Model – regression

relaonship based on past me series

values to predict the future me series

values

Seasonality without Trend

LinReg in Excel

Seasonality with Trend

ClassDecomp in Excel

Seasonal Index

> 1 = above average sales

≈ 1 = close to average sales

< 1 = below average sales

Seasonally Adjusted Forecast

= Forecast * Seasonal Index

Two Sigma Control Chart – errors

should be distributed about zero to

use best model; limits of ±2√(MSE)

Decision Analysis – develop opmal

strategy when a decision maker is

faced with several decision

alternaves and an uncertain or risk

"lled paern of future events

Opmisc Approach - leads to

choosing the decision alternave

corresponding to the largest payo&

Maximax in Excel

Conservave Approach – leads to

choosing the decision alternave that

maximizes the minimum payo&

Maximini in Excel

Regret Opportunity Loss – leads to

choosing the decision alternave that

minimizes the maximum regret / loss

Minimax Regret in Excel

Prior Probabilies – probabilies of

the states of nature prior to obtaining

sampling informaon; preliminary

assessments for states of nature

EV in Excel

Expected Value (EV) – the weighted

averages of the payo&s; weights are

states of nature

At Chance Nodes – mulplying the

payo& at the end of each brand by

the corresponding branch

probabilies

At Decision Nodes – select the

branch that leads to the best

expected value