Forecasting

Chapter 3
 Forecasting
 It is the art and science of  Features Common to All Forecasts
predicting future event.
 Business Forecasting – is an 1. Techniques assume some underlying causal
estimate or prediction of future system that existed in the past will persist into
the future
developments in business such as
2. Forecasts are not perfect
sales, expenditures, revenues etc.
3. Forecasts for groups of items are more
accurate than those for individual items
 One of the most important aspect in 4. Forecast accuracy decreases as the
corporate planning. forecasting horizon increases

 Demand Forecasting – is a forecast

that predicts company sales.
 Factors to consider:

Elements of a Good Forecast Steps in the Forecasting Process

The forecast 1. Determine the purpose of the forecast

1. should be timely 2. Establish a time horizon
2. should be accurate 3. Obtain, clean, and analyze appropriate
3. should be reliable data
4. should be expressed in meaningful units 4. Select a forecasting technique
5. should be in writing 5. Make the forecast
6. technique should be simple to 6. Monitor the forecast errors
understand and use
7. should be cost-effective
 Forecast Accuracy Metrics

MAD 
 Actual t  Forecast t Mean Absolute Deviation
n MAD weights all errors evenly

 Actual t  Forecast t 
2
Mean Squared Error
MSE 
n 1 MSE weights errors according to
their squared values
Actual t  Forecast t
 Actual t
 100
MAPE  Mean Absolute Percent Error
n
MAPE weights errors according to
relative error
Forecast Error Calculation
Forecasting Methods:
Forecasting
Method

Quantitative Qualitative
Methods Methods

Time Series Causal Delphi

Methods Methods Methods

Jury of
Moving Regression
Executive
Averages Analysis
Opinions

Exponential Multiple Sales Force

Smoothing Regression Composite

Consumer
Trend
Market
Projections
Survey
 Forecasting Approaches:
B. Associative Model – forecasting technique that uses explanatory variables to
1. Qualitative Forecasting product future demand.
1. Executives Opinion – often used as a part of long-range planning and
– Forecasts that use subjective inputs new product development
such as opinions from consumer 2. Sales Force Composite – is a good source of information because of its
direct contrast with consumers
surveys, sales staff, managers, 3. Consumer Surveys – It can tap information that might not be available
executives, and experts elsewhere
4. Outside Opinion – this may concern advice on political or economic
conditions is a foreign country or some other aspects of interest with
which an organization lacks familiarity
A. Judgmental Forecasts – rely on analysis of subjective inputs 5. Opinions of Managers and Staff – At times, a manager may solicit from
obtained from various sources, such as consumer surveys, the sales a number of other managers and/or staff. The Delphi Method is useful
in this regard.
staff, managers, and executives, panels of experts etc.
 Forecasting Approaches:
2. Quantitative Forecast – time series – a time ordered 1. Naïve Forecast – a forecast for any period equals the
sequence of observations taken at regular intervals over previous period’s actual value.
time. Plotting the data and visually examining the plot. 2. Moving Average – making use of the most recent data to get
• Assume that future values of the time-series can be the forecast.
estimated from past values of the time-series
- Trend – long term upward of downward movement in a data
3. Weighted Moving Average – weight can be used to place
1. Seasonal – short-term regular variations related to weather more emphasis in recent values, when there is a trend or
or other factors pattern. This makes the technique more responsive to
2. Cyclical – wavelike variation lasting more than one year changes since more recent period may be more heavily
3. Irregular – variations caused by unusual circumstances not weighted.
by reflective or typical behavior
4. Random – residual variations after all other behaviors are 4. Exponential Smoothing – Each new forecast is based on the
accounted for. previous forecast plus a percentage of the difference between
that forecast and the actual value of the series at that point
 Forecast based on Historical Data – a technique that 5. Simple Linear Regression – the simple and most widely
depends on uncovering relationships between variables that
can be used to predict future values of one of them.
used form of regression involves a linear relationship
 Averaging Technique – generate forecasts that reflect
between two variables. The objective in linear regression is
recent values of a time-series to obtain an equation of a straight line that minimizes the
 Trend Line – associative to series of movements from a sum of equation vertical deviations of points around the line.
straight line.
Trend:
 Illustrative Examples:
Averaging Techniques: Compute a three-period moving average forecast given
the following demand for cars for the last five periods.

 Moving Average Week Demand

1 70
2 80
Moving average =  Demand in previous n periods
n 3 65
Where: 80 4 90
n – is the number of period in the moving 85 5 85
average 6 ?
83
 Illustrative Examples:
 Weighted Moving Average Compute for week 6 forecast using the weights: 50%,
30% and 20% respectively given the following demand
(weight for period n)(demand in period n) for cars for the last five periods.
 Weights

(weight for period n)(demand in period n)

83

= 83(.5) + 85(.3) + 90(.2) 85

84
 Illustrative Examples:
 Exponential Smoothing 1. In January, a demand for 200 units of Toyota car model
“Vios” for February was predicted by a car dealer.
 New forecast = Last Period’s Forecast +  (Last Period’s Actual February demand was 250 cars. Forecast the
Actual Demand – Last Period’s Forecast)
March demand using a smoothing constant of  = 0.30.
 Where:  represents the value of a weighing factor – smoothing factor –
value is 0 and 1. New Forecast: 200 + 0.30(250-200) = 215 cars
Ft = Ft – 1 +  [At – 1 – Ft - 1] 2. Use exponential smoothing model to develop a series of
Where: forecast for the following data and compute.
Ft – the new forecast or forecast for period
a. Use a smoothing
PERIOD
factor of 0.20
ACTUAL DEMAND
Ft-1 – the previous forecast or forecast for period t-1 1 20
 - smoothing constant 2 35
3 46
At-1 – actual demand or sales for period t-1 4 40
5 50
6 55
 The smoothing constant, , represents percentage of the 7 45
8 ?
forecast error. Each new forecast is equal to the previous
forecast plus a percentage of the previous
Period Actual Forecast Solution Answer
Demand
F3 = 20 + 0.20(35 – 20) 23
1 20 -
F4 = 23 + 0.20(46 – 23) 27.60
2 35 20
F5 = 27.60 + 0.20(40 – 27.60) 30.08
3 46 23
F6 = 30.08 + 0.20(50 – 30.08) 34.06
4 40 27.60
F7 = 34.06 + 0.20(55 – 34.06) 38.25
5 50 30.08
F8 = 38.25 + 0.20(45 – 38.25) 39.60
6 55 34.06
7 45 38.25
8 ? 39.60
 Illustrative Examples:
 Trend Line Forecast – Least Square Method: A straight line that minimizes
the sum of the vertical differences from the line to each of the data points. The
 Linear trend equation: 1. Given: DVD Sales of ABC Marketing
Tt = a + btx
Where: A. Determine
Year
the forecast
DVD Sales time Series
sales for
Sales (Units/1, 000)
tx – independent variable
Tt – computed value of the variable to be predicted (dependent
2010 and 2011
2001 3
variable)
a - intercept of the trend line (Y-axis intercept)
2002 4.5
b - slope of the trend line 2003 4.8
the coefficient of the line a and b can be computed using two equations: 2004 3.7
2005 4.6
b =  ty - ny
t2 – n2
2006 5
a = 2007 4
2008 5
where:
2009 6
n – number of data points or obervations
Y – values of the dependent variables
2010 ?
- Average of the values of the Y’s 2011 ?
t – values of the independent variables
- Average of the values of the X’s
 Illustrative Examples:
DVD Sales time Series
Year Period Sales (1, 000) ty
Solutions:
t Y

2001 1 3 1 3
2002 2 4.5 4 9
2003 3 4.8 9 14.4
2004 4 3.7 16 14.8
2005 5 4.6 25 23
2006 6 5 36 30
2007 7 4 49 28
2008 8 5 64 40
2009 9 6 81 54
= 45 = 40.6
The Trend Equation will be:
 Illustrative Examples:
Sales forecast for 2010 Sales forecast for 2011
 Illustrative Examples:
 Regression Analysis – It is a statistical technique used to
develop a mathematical equation showing how variables are
related. It is a forecasting technique that uses the least square 1. Dumlao Construction Firm renovates homes in
approach on one or more independent variables. Marilao, Bulacan. Over time, the business found that
 Formula its Peso volume renovation work is dependent in the
Marilao Bulacan payroll. The data for Dumlao’s
Where: revenue and Ythe amount of money earned
X by wage
Dumlao’s Sales (P100, Payroll (P1, 000, 000)
X – the independent variable earners in Marilao
000) Bulacan for the past 5 years are
– value of the dependent variable shown below:
a - Y-axis intercept 3.0 2
b – scope of the regression line 2.0 3
the coefficient of the line a and b can be computed using 3.5 6
two equations:
2.0 5
3.0 4
2

a = -b
 Illustrative Examples:
Using Least Squares Regression Analysis:
Sales Payroll
Y X

3.0 2 4 6.0
2.0 3 9 6.0 0
3.5 6 36 21.0
2.0 5 25 10.0
.30
3.0 4 16 12.0
= 13.5 = 20 = 90 = 55
The estimated regression equation is:
 Illustrative Examples:
 If Dumlao Construction wishes to have a payroll of
Php 5, 500, 000 next year, an estimated sales for
Dumlao Construction is:

Sales (P100, 000) = 2.30 + 0.1(Payroll)

= 2.30 + 0.1(5, 500, 000)
= 2.30 + 0.55
= 2.85
Sales = Php 2,850,000.00
 Practice:
1. Use quantitative forecast methods for the data
shown below:

Period Observat
ion Compute using:
1 24 1. Naïve Method
2 34 2. Three Period Moving Average
3. 4 period weighted moving average
3 36 4. Exponential Smoothing with .30 factor
4 37 5. Least Square Method
5 41
6 44
7 45
8 ?