Session 5

Forecasting 1
Fall 2021
Department of Business Administration
UC3M
Professor Gemma Berenguer
 Agenda
• Forecasting 1

Operations Management
2

Gemma Berenguer
Characteristics of Forecasts
1. Forecasts are always inaccurate
2. Long-term forecasts are usually less accurate than short-
term forecasts

Gemma Berenguer
Operations Management
3. Aggregate forecasts are usually more accurate than
disaggregate forecasts
4. In general, the farther up the supply chain a company is,
the greater is the distortion of information it receives

3
Components and Methods
• Companies must identify the factors that
influence future demand and then ascertain the
relationship between these factors and future

Gemma Berenguer
Operations Management
demand

4
Components and Methods
1. Time Series
• Use historical demand only
• Best with stable demand

Gemma Berenguer
Operations Management
2. Qualitative
• Primarily subjective
• Rely on judgment
3. Causal
• Relationship between demand and some other factor
4. Simulation
• Imitate consumer choices that give rise to demand 5
Components of an Observation
Observed demand (O) = systematic component (S)
+ random component (R)

• Systematic component – expected value of demand

Gemma Berenguer
Operations Management
− Level (current deseasonalized demand)
− Trend (growth or decline in demand)
− Seasonality (predictable seasonal fluctuation)
• Random component – part of forecast that deviates
from systematic component
• Forecast error – difference between forecast and actual
demand 6
Behaviors
• Level

Demand
No Pattern

Gemma Berenguer
Operations Management
• Trend Time

Linear (default) or Nonlinear

• Seasonality
Repetition at Fixed Intervals

7
 Components of Demand
Trend
component
Demand for product or service

Seasonal peaks

Actual demand
line

Average demand
over 4 years

Random variation
| | | |
1 2 3 4
Time (years)
Figure 4.1
Time Series Models

Forecasting Method Applicability

Gemma Berenguer
Operations Management
Moving average No trend or seasonality

Exponential smoothing No trend or seasonality

Holt’s model Trend but no seasonality

Winter’s model Trend and seasonality (not in this course)

9
Steps in Forecasting
• Initialize
• Compute initial estimates of level (L0), trend (T0), and
seasonal factors (S1,…,Sp), p: periodicity of seasonal
demand

Gemma Berenguer
Operations Management
• Forecast
• Forecast demand for period t + 1 (Ft+1)
• Estimate error
• Compute error Et+1 = Ft+1 – Dt+1
• Revise estimates (adaptive)
• Modify the estimates of level (Lt+1), trend (Tt+1), and
seasonal factor (St+p+1) 10
Moving Average
❑Used when demand has no observable trend or seasonality
❑Systematic component of demand = level

❑The level in period t is the average demand over the last N

Gemma Berenguer
Operations Management
periods
Lt = (Dt + Dt-1 + … + Dt–N+1) / N

❑The forecast in period t+1 is

Ft+1 = Lt

❑After observing the demand for period t + 1, revise the

estimates 11
Lt+1 = (Dt+1 + Dt + … + Dt-N+2) / N, Ft+2 = Lt+1
Moving Average Example
• A supermarket has experienced weekly demand of milk of D1 =
120, D2 = 127, D3 = 114, and D4 = 122 gallons over the past four
weeks:

Gemma Berenguer
Operations Management
• Forecast demand for Period 5 using a four-period moving average
120,75

• What is the forecast error if demand in Period 5 turns out to be 125

gallons?
- 4,25

12
Moving Average Example
❑Forecast demand for Period 5

❑Error if demand in Period 5 = 125 gallons

Gemma Berenguer
Operations Management
❑ Next, step revised demand
Comments on MA
• What historical data (how many past periods) does this
method use?
N

Gemma Berenguer
Operations Management
• What weight is given to old and new observations?
The same

• What is the impact of N?

As N increases, the model is more stable, but less responsive to the most recently observed demand

14
Exponential Smoothing
❑Used when demand has no observable trend or seasonality
❑Systematic component of demand = level
❑The level in period t is

Gemma Berenguer
Operations Management
Lt = a Dt +(1– a )Ft , L0 = D0
❑The forecast in period t+1 is

Ft+1 = Lt
❑After observing the demand for period t + 1, revise the estimates

Ft+2 = Lt+1 = a Dt+1 +(1– a )Ft+1

15
Example Exponential
Smoothing
• A supermarket has experienced weekly demand of milk of D1 = 120,
D2 = 127, D3 = 114, and D4 = 122 gallons over the past four weeks
• Forecast demand for Period 5 using exponential smoothing with a = 0.1

Gemma Berenguer
Operations Management
• What is the forecast error if demand in Period 5 turns out to be 125
gallons?

16
Comments on ES
• What historical data (how many past periods) does this
method use?
ALL

Gemma Berenguer
Operations Management
• What weight is given to old and new observations?
More weight to the most recent one

• What is the impact of α?

As "alpha" increases, the process is more responsive to changes in recent demand

17
Trend-Corrected Exponential
Smoothing (Holt’s Model)
• Appropriate when the demand is assumed to have a level and
trend in the systematic component of demand but no seasonality
Systematic component of demand = level + trend

Gemma Berenguer
Operations Management
• Obtain initial estimate of level and trend by running a linear
regression
Dt = at + b, T0 = a, L0 = b
• In Period t the forecast for future periods is
Ft+1 = Lt + Tt and Ft+n = Lt + nTt
• After observing demand of time t+1, revise

Lt+1 = aDt+1 + (1 – a)(Lt + Tt), Tt+1 = b(Lt+1 – Lt) + (1 – b)Tt 18

Example Holt’s Model
• MP3 player demand
D1 = 8,415, D2 = 8,732, D3 = 9,014,
D4 = 9,808, D5 = 10,413, D6 = 11,961
a = 0.1, b = 0.2

Gemma Berenguer
Operations Management
• Using regression analysis we obtain
L0 = 7,367(intercept) and T0 = 673(slope)
• Forecast for period 1 is
F1 = L0 + T0 = 7,367 + 673 = 8,040

19
Example Holt’s Model
• What is the forecast of period 7?

Gemma Berenguer
Operations Management
20
Measures of forecast error
• Error E = F – D
t t t
𝑛
• Bias 𝑏𝑖𝑎𝑠𝑛 = ෍ 𝐸𝑡

Gemma Berenguer
Operations Management
𝑡=1
• Tracking Signal (TS) = Bias/MAD <= |6|
𝑛
1
• Mean absolute deviation (MAD) 𝑀𝐴𝐷𝑛 = 𝑛 ෍ 𝐴𝑡 , 𝐴𝑡 = |𝐸𝑡 |
𝑛 𝑡=1
1
• Mean squared error (MSE) 𝑀𝑆𝐸𝑛 = ෍ 𝐸𝑡2
𝑛
𝑡=1 𝐸𝑡
σ𝑛𝑡=1 100
• Mean absolute percent error (MAPE) 𝐷𝑡
𝑀𝐴𝑃𝐸𝑛 = 21
𝑛
Criteria for selecting method
• Plot your historical demand and look for patterns: level, trend,
and seasonality
• Forecast errors provide important information for choosing
the best forecasting method

Gemma Berenguer
Operations Management
• Forecast errors can also be used to select the best values for
the parameters needed for the method
• Criteria used:
• Minimizing bias;
• Minimizing MAPE, MAD or MSE;
• Meeting managerial expectations

22
Several pointers for quantitative
methods forecasting
• Don’t forget to distinguish between sales data and demand
data
• Bias is the worst kind of forecast error; strive for zero bias

Gemma Berenguer
Operations Management
• Whenever possible, forecast at aggregate levels. Forecast in
detail only where necessary

23