Professional Documents
Culture Documents
Finance Operations
Cash flow and Schedules, MRP, workloads
funding
HR Product Design
Hiring/ recruiting/ New products and services
training
FORECASTS
Seasonality
short-term regular
variations in data Random
Variations
Cycles
Caused by chance
wavelike variations
of long-term
TREND
The systematic
increase or
decrease in the
mean of the series
over time
SEASONAL
A repeatable
pattern of increases
or decreases in
demand, depending
on the time of day,
week, month, or
season.
CYCLICAL
The less predictable
gradual increases or
decreases in demand
over longer periods
of time (years or
decades).
NAÏVE FORECASTS
• Uses a single previous value of a time series as the basis of a
forecast.
• Virtually no cost
• Data analysis is nonexistent
• Easily understandable
• Cannot provide high accuracy
• If it were true, future will always be the same as the past
F(4) = 411
Time Series Models: Variations
What is random and what is not?
• Historical data contain random variations or noise
• Random variations are caused by relatively unimportant
factors.
• What is random? Can we not study everything to negligible
detail? “God does not roll dices” –A.E.
• The objective is to remove all randomness and have real
variations.
• Minor variations are random and large ones are real.
TECHNIQUES FOR AVERAGING
Moving
Averages
Exponential
Smoothing
Weighted
Moving
Averages
MOVING AVERAGES
• Involves calculating the average demand for the n most recent
time periods and using it as the forecast for future time periods
ILLUSTRATIVE PROBLEM 2
Using the three-week forecasting method, what is the
forecast for Week 4?
F(4) = (400+380+411) / 3
F(4) = 397
WEIGHTED MOVING AVERAGES
• Each historical demand in the average can have its own
weight
• The sum of the weights equals 1.0.
• For example, in a three-period weighted moving average
model, the most recent period might be assigned a weight of
0.50, the second most recent might be weighted 0.30, and the
third most recent might be weighted 0.20.
ILLUSTRATIVE PROBLEM 3
The Polish General’s Pizza Parlor is a small restaurant catering to patrons with a
taste for European pizza. One of its specialties is Polish Prize pizza. The manager
must forecast weekly demand for these special pizzas so that he can order pizza
shells weekly. Recently, demand has been as follows:
F(@Jun23) = 0.5*52 +
0.3*65 + 0.2*50 = 55.5
F(@Jun30) = 0.5*56 +
Forecast June 23, 30 and July 7 by using
0.3*52 + 0.2*65 = 56.6
the weighted moving average method with
n = 3 and weights of 0.50, 0.30, and 0.20,
F(@Jul7) = 0.5*55 + 0.3*56
with 0.50 applying to the most recent + 0.2*52 = 54.7
demand.
MA vs. WMA
Moving Weighted Moving
Averages Averages
Advantage: Advantage:
Easy to compute and easy Allows you to emphasize recent
to understand demand over earlier demand
Disadvantage: The forecast will be more
All values in the average responsive to changes in the
are weighted equally underlying average of the
demand series than the simple
moving average forecast.
EXPONENTIAL SMOOTHING
• A sophisticated weighted moving average method that calculates the
average of a time series by implicitly giving recent demands more weight
than earlier demands, all the way back to the first period in the history file.
• Most frequently used formal forecasting method because of its simplicity
and the small amount of data needed to support it.
• Requires only three items of data: (1) the last period’s forecast; (2) the
actual demand for this period; and (3) a smoothing parameter, alpha (a),
which has a value between 0 and 1.0.
ILLUSTRATIVE PROBLEM 4
Forecast using the exponential smoothing with a smoothing
constant value of 0.10, and forecast value for week 3 of 390
The number of patient arrivals for the past
three weeks were as follows:
y = a + bx
Where
y = predicted (dependent) variable
x = predictor (independent) variable
b = slope of the line
a = value of y when x = 0 (the height of
line at the y intercept)
LINEAR REGRESSION
• Correlation (r) between variables: The strength
and direction of relationships between two
variables
• 1.00 means changes in one variable are always
matched by changes in the other, vice versa.
• A correlation close to zero means little linear
relationship
• The square of the correlation coefficient provides
a measure of the percentage of variability in the
values of y that is explained by the independent
variable.(80% or more: the independent variable
is a good predictor of the values of dependent
variable)
FORECAST ACCURACY
1. Measurement is the first step to improve
an activity
a. What value of smoothing constant is good?
| A F | t t
MAD t 1
n
n n
(A F )t t
2
(A F ) t t
2
MSE t 1
t 1
n 1 n
n
A F t t
Tracking Signal t 1
MAD
Estimate of (forecast error) standard deviation s MSE
Statistics says : MSE is the unbiased estimator for the variance of forecast error.
ILLUSTRATIVE PROBLEM 5
The monthly demand for units
manufactured by the Acme Rocket Month Units
Company has been as follows. May 100
Use the exponential smoothing method to
Jun 80
forecast the number of units for June to
January. The initial forecast for May was Jul 110
105 units; a = 0.2. b. Calculate the Aug 115
absolute percentage error for each month
from June through December and the Sept 105
MAD and MAPE of forecast error as of the Oct 110
end of December. c. Calculate the tracking Nov 125
signal as of the end of December. What
can you say about the performance of Dec 120
your forecasting method?
ILLUSTRATIVE PROBLEM 5
Month Actual Forecast Error AD SE %
May 100 105 A-F |E| E^2 E/A
Jun 80 100*0.2+105*0.8 104 -24 24 576 30%
Jul 110 80*0.2+104*0.8 99.2 10.8 10.8 116.64 9.8%
Aug 115 110*0.2+99.2*0.8 101.4 13.6 13.6 184.96 11.8%
Sept 105 115*0.2+101.4*0.8 104.1 0.9 0.9 0.81 0.9%
Oct 110 105*0.2+104.1*0.8 104.3 5.7 5.7 32.49 5.2%
Nov 125 110*0.2+104.3*0.8 105.4 19.6 19.6 384.16 15.7%
Dec 120 125*0.2+105.4*0.8 109.3 10.7 10.7 114.49 8.9%
37.3 85.3 1,409.55 82.3%
5.3 12.2 201.4 11.8%
USE FOR MAD AND MSE