Course Code : MN 3197

Course Title: Operation Management

Tutorial -8
24-11-2025

Chapter 18
Forecasting
Required Formulae
 Required Formulae
❑ The equations to compute the forecast including trend (FIT) are,
Ft= FITt−1 + α (At−1 - FITt−1 )
Tt = Tt -1 + δ (Ft - FITt−1 )
FITt = Ft + Tt
Ft= The exponentially smoothed forecast that does not include trend for period t
Tt= The exponentially smoothed trend for period t
FITt = The forecast including trend for period t
FITt−1= The forecast including trend made for the prior period
At−1 = The actual demand for the prior period
α = Smoothing constant (alpha)
δ= Smoothing constant (delta)
Required Formulae
 Decomposition of a Time Series
A time series may contain one or many elements :

1. Trend
Definition: The persistent, long-term upward or downward movement in the data, showing
the overall direction over an extended period.
Example: Temperature Rise due to global warming
2. Seasonal
Definition: A predictable and repeating pattern that occurs over a fixed, known period
(within a single year, month, or week).
Example: Shopping habit during festivals
3. Cyclical
Definition: Long-term fluctuations related to economic cycles, lasting several years, with no
fixed or predictable period.
Example: Population growth , Product life cycles
 Decomposition of a Time Series
4. Autocorrelation
Definition: The "memory" of a series; the degree to which its current value is correlated with
its own immediate past values.
Example : Daily Market Price
5. Random (Irregular or Noise)
Definition: The completely unpredictable and erratic component that remains after all other
systematic patterns have been accounted for.
Example : Production disturbance due to power shortage
Additive vs. Multiplicative Seasonal Variation
When a time series has both a trend and a seasonal pattern, we need to define how these
components interact. The two primary models are Additive and Multiplicative.
1. Additive Model
Formula: Y(t) = Trend(t) + Seasonal(t)
Concept: The seasonal fluctuations are constant in magnitude, regardless of the trend level.
The peaks and troughs of the seasonality are roughly the same size throughout the
series.
When to Use: When the seasonal variation appears to be constant over time.
Example: Monthly Hotel Room Nights in a Steady City
▪ The trend is relatively flat (e.g., around 10,000 room nights per month).
▪ The seasonal effect is a consistent boost of +2,000 rooms every July and a drop of 1,500
rooms every February
Additive vs. Multiplicative Seasonal Variation
2. Multiplicative Model
Formula: Y(t) = Trend(t) x Seasonal(t)
Concept: The seasonal fluctuations are proportional to the trend level. As the trend
increases, the size of the seasonal swings also increases.
When to Use: When the seasonal variation appears to grow (or shrink) with the level of the
series. This is very common in business and economic data.
Example: Sales of a Successful New Smartphone
Year 1 Trend: Sales are around 1 million units per quarter.
Year 1 Seasonal: Q4 (holiday season) sees a 50% increase.
So, sales in Year 1 Q4 are = 1 million * 1.5 = 1.5 million.
Year 2 Trend: The product is more popular, with a trend of 2 million units per quarter.
Year 2 Seasonal: The Q4 seasonal effect is still a 50% increase.
Sales in Year 2 Q4 are 2 million * 1.5 = 3 million.
 What is a Seasonal Factor?
Definition: A seasonal factor is a numerical value that represents how a particular season
(month, quarter, etc.) compares to the average season. It measures the typical percentage
deviation from the average.

Seasonal Factor = (Season Average) ÷ (Overall Average)

 Practice Problem-1
Sunrise Baking Company markets doughnuts through a chain of food stores. It has been
experiencing overproduction and underproduction because of forecasting errors. The
following data are its demand in dozens of doughnuts for the past four weeks. Doughnuts are
made for the following day; for example, Sunday’s doughnut production is for Monday’s sales,
Monday’s production is for Tuesday’s sales, and so forth. The bakery is closed Saturday, so
Friday’s production must satisfy demand for both Saturday and Sunday.
Make a forecast for this week based on the following:
a. Daily, using a simple four-week moving average.
b. Daily, using a weighted moving average with weights of 0.40, 0.30, 0.20, and 0.10 (most
recent to oldest week).
 Practice Problem-1
c. Sunrise is also planning its purchases of ingredients for bread production. If bread demand
had been forecast for last week at 22,000 loaves and only 21,000 loaves were actually
demanded, what would Sunrise’s forecast be for this week using exponential smoothing with α
= 0.10?
d. Suppose, with the forecast made in c, this week’s demand actually turns out to be 22,500.
What would the new forecast be for the next week?
 Solve
a. Simple four-week moving average forecast
Monday Friday (must cover Sat + Sun demand)
= (2200 + 2400 + 2300 + 2400) / 4 = (1900 + 1800 + 2100 + 2000) / 4
= 9300 / 4 = 2325 = 7800 / 4 = 1950
Tuesday Sunday
= (2000 + 2100 + 2200 + 2200) / 4 = (2800 + 2700 + 3000 + 2900) / 4
= 8500 / 4 = 2125 = 11400 / 4 = 2850
Wednesday
= (2300 + 2400 + 2300 + 2500) / 4
= 9500 / 4 = 2375
Thursday
= (1800 + 1900 + 1800 + 2000) / 4
= 7500 / 4 = 1875
 Solve
b. We multiply last week (0.40), 2 weeks ago (0.30), 3 weeks ago (0.20), 4 weeks ago (0.10).

Monday Friday
= 0.4×2400 + 0.3×2300 + 0.2×2400 + 0.1×2200 = 0.4×2000 + 0.3×2100 + 0.2×1800 + 0.1×1900
= 960 + 690 + 480 + 220 = 2350 = 800 + 630 + 360 + 190 = 1980
Tuesday Sunday
= 0.4×2200 + 0.3×2200 + 0.2×2100 + 0.1×2000 = 0.4×2900 + 0.3×3000 + 0.2×2700 + 0.1×2800
= 880 + 660 + 420 + 200 = 2160 = 1160 + 900 + 540 + 280 = 2880
Wednesday
= 0.4×2500 + 0.3×2300 + 0.2×2400 + 0.1×2300
= 1000 + 690 + 480 + 230 = 2400
Thursday
= 0.4×2000 + 0.3×1800 + 0.2×1900 + 0.1×1800
= 800 + 540 + 380 + 180 = 1900
 Solve
c. Exponential smoothing for bread demand

Formula, Ft =Ft-1 + α ( At-1 – Ft-1 )

Ft-1 =22000, At-1 =21000 α=0.10

So, Ft = 22000+0.10×(21000−22000) = 22000−100=21900

d. Update with new data

Ft=21900, At =22500, α=0.10

So, Ft+1 = 21900+0.10×(22500−21900) = 21900+60= 21960

 Practice Problem-2
The following table contains the demand from the last 10 months:

Month Actual Demand

1 31
2 34
3 33
4 35
5 37
6 36
7 38
8 40
9 40
10 41
 Practice Problem-2
a. Calculate the single exponential smoothing forecast for these data using an α of 0.30 and
an initial forecast ( F1 ) of 31.
b. Calculate the exponential smoothing with trend forecast for these data using an α of 0.30,
a δ of 0.30, an initial trend forecast ( T1 ) of 1.00, and an initial exponentially smoothed
forecast (F1 ) of 30.00.
c. Calculate the mean absolute deviation (MAD) for each forecast. Which is best?
 Solve
a. Exponential smoothing Formula, Ft =Ft-1 + α ( At-1 – Ft-1 )

For First month, F1 = 31 ( It’s given, as there is no At-1 (Actual Sale) before 1st month)

For Second month , For Fourth month ,

Ft-1 = F1 =31, At-1 =31 α=0.30 Ft-1 = F3 =31.9, At-1 =33 α=0.30
So, F2 = 31+0.30×(31-31) = 31 So, F3 = 31.9+0.30×(33-31.9) =32.23
For Third month , For Fifth month ,
Ft-1 = F2 =31, At-1 =34 α=0.30 Ft-1 = F4 =32.23, At-1 =35 α=0.30
So, F3 = 31+0.30×(34-31) =31.9 So, F4 = 32.23+0.30×(35-32.23) =33.06
This same calculation need to be continued for all the months
 Solve
Month Actual Demand (At) Exponential Smoothing
Forecast (Ft)
1 31 31
2 34 31
3 33 31.9
4 35 32.23
5 37 33.06
6 36 34.24
7 38 34.77
8 40 35.74
9 40 37.02
10 41 37.91
 Solve
b. The equations to compute the forecast including trend (FIT) are,
Ft= FITt−1 + α (At−1 - FITt−1 )
Tt = Tt -1 + δ (Ft - FITt−1 )
FITt = Ft + Tt
For First Month, Ft = F1 = 30,( It’s given, as there is no At-1 (Actual Sale) and FITt−1 before 1st month)
Tt = T1 = 1 , ( It’s given, as there is no Tt -1 and FITt−1 before 1st month)
FITt = FIT1 = Ft + Tt = 30 + 1 = 31
For Second Month, FITt−1 = FIT1 = 31, At−1 =A1 = 31 , Tt -1 = T1 = 1 α =0.30, δ =0.30
Ft = F2 = FITt−1 + α (At−1 - FITt−1 ) = FIT1 + α (A1 – FIT1 ) = 31 + 0.30 x ( 31-31) = 31 + 0.30 x 0 = 31
Tt = T2 = Tt -1 + δ (Ft - FITt−1 ) = T1 + δ (F2 - FIT1 ) = 1 + 0.30 x (31 -31) =1 + 0.30 x 0 = 1
FITt = FIT2 = Ft + Tt = F2 + T2 = 31 + 1 = 32
 Solve
For Third Month, FITt−1 = FIT2 = 32, At−1 = A2 = 34 , Tt -1 = T2 = 1 α =0.30, δ =0.30
Ft = F3 = FITt−1 + α (At−1 - FITt−1 ) = FIT2 + α (A2 – FIT2 ) = 32 + 0.30 x ( 34-32) = 32 + 0.30 x 2 = 32.6
Tt = T3 = Tt -1 + δ (Ft - FITt−1 ) = T2 + δ (F3 - FIT2 ) = 1 + 0.30 x (32.6 -32) =1 + 0.30 x 0.6 = 1.18
FITt = FIT3 = Ft + Tt = F3 + T3 = 32.6 + 1.18 = 33.78

For Four Month, FITt−1 = FIT3 = 33.78, At−1 = A3 = 33 , Tt -1 = T3 = 1.18 α =0.30, δ =0.30
Ft = F4 = FITt−1 + α (At−1 - FITt−1 ) = FIT3 + α (A3 – FIT3 ) = 33.78 + 0.30 x ( 33-33.78) = 33.55
Tt = T4 = Tt -1 + δ (Ft - FITt−1 ) = T3 + δ (F4 - FIT3 ) = 1.18 + 0.30 x (33.55 -33.78) =1 + 0.30 x 0.6 = 1.11
FITt = FIT4 = Ft + Tt = F3 + T3 = 33.55 + 1.11 = 34.66
This same calculation need to be continued for all
the months
 Solve
Month Actual Exponential Smoothing The exponentially smoothed The forecast including trend
Demand (At) Forecast Without Trend (Ft) trend for period t (Tt) for period (FITt)

1 31 31 1 31
2 34 31 1 32
3 33 31.9 1.18 33.78
4 35 32.23 1.11 34.66
5 37 33.06 1.14 35.90
6 36 34.34 1.24 37.47
7 38 34.77 1.11 38.14
8 40 35.74 1.10 39.19
9 40 37.02 1.17 40.60
10 41 37.91 1.11 41.54
 Solve
C. Mean absolute deviation (MAD) for Exponential Smoothing Forecast
Month Actual Demand Exponential Absolute
(At) Smoothing Deviation
Forecast (Ft) | At - Ft |
1 31 31
2 34 31 |34-31|=3
MAD = 26.13 / 9 = 2.90
3 33 31.9 1.1
4 35 32.23 2.77
5 37 33.06 3.94
6 36 34.34 1.76
7 38 34.77 3.23
Why n= 9?
8 40 35.74 4.26
Because error or deviation
9 40 37.02 2.98
available for 9 period, no error
10 41 37.91 3.09 calculation for first period.
26.13
 Solve
C. Mean absolute deviation (MAD) for Exponential smoothing with trend forecast
Month Actual Demand The forecast Absolute
(At) including trend Deviation
for period (FITt) | At - FITt |
1 31 31
2 34 32 |34-32|=2
MAD = 7.77 / 9 = 0.86
3 33 33.78 0.78
4 35 34.67 0.33
5 37 35.90 1.10
6 36 37.47 1.47
7 38 38.14 0.14
Based upon the MAD of each forecast,
8 40 39.19 0.81
the exponential smoothing with trend is
9 40 40.60 0.60
the better forecasting model.
10 41 41.54 0.54
7.77
 Past Question – October 2023

1) Compute the overall tracking signal using the mean absolute deviation and the running
sum of forecast errors.
2) Calculate the tracking signals for each forecast and plot them on a graph.
3) Using your results in part 1 and part 2, evaluate the accuracy of the forecasting model.
 Solve
1) Cumulative
Actual Forecast Actual Absolute
Month Deviation
Demand Demand Deviation Deviation
(RSFE)
1 842 800 42 42 42

2 917 900 17 59 17

3 937 975 -38 21 38

4 1008 1050 -42 -21 42

5 1120 1150 -30 -51 30

Sum= 169
 Solve
2) Cumulative
Actual Forecast Actual Absolute Sum of
Month
Demand Demand Deviation
Deviation
Deviation Abs. Dev.
MAD TS
(RSFE)
842-800 42/42
1 842 800 42 |42|=42 42 42/1=42
=42 =1
917-900 42+17 42+17 59/2 59/29.5
2 917 900 |17|=17
=17 = 59 =59 =29.5 =2
59-38 59+38
3 937 975 -38 |-38|=38 32.33 0.65
=21 =97

4 1008 1050 -42 -21 42 139 34.75 -0.60

5 1120 1150 -30 -51 30 169 33.8 -1.51

 Solve
3) Based on the tracking signal analysis:
While the current tracking signal of 1.51 falls
within acceptable limits (±4), concerning
trends are emerging. The tracking signal
shows a consistent downward trajectory and
MAD values are increasing, suggesting
potential bias in the forecasting model.
Although based on limited data, these patterns
indicate the model may be systematically
over-forecasting. Continuous monitoring is
Tracking Signal plotted against each month forecast essential to ensure the tracking signal doesn't
continue declining and that actual demand
doesn't persistently fall below forecasts.
 Past Question – May 2025
Tailored by Amira is a renowned bespoke tailoring studio specialising in crafting exquisite
wedding attire, including bridal gowns, suits, and traditional garments. Amira and her
team use only the finest, sustainably sourced fabrics to ensure every piece is both elegant
and environmentally responsible.
The amount of fabric required per wedding varies, as some clients order only a single
gown while others request multiple outfits for the entire wedding party. However, over
the years, Amira has observed that the total fabric usage in her studio is closely
influenced by the number of wedding licences issued annually in the region. Analysis of
historic data confirms a linear relationship between the number of licences issued and
total fabric usage. Based on this, Amira has decided to use this linear relationship to
forecast her fabric needs for the coming year.
 Past Question – May 2025
For this analysis, Amira is asking you to base your calculations on the following data:

1) Assuming the linear relationship between the number of wedding licenses and fabric
usage continues, use the data provided and linear regression analysis to estimate the
amount of fabric (in m2 ) that Tailored by Amira will need in 2025.
 Solve

Number of Wedding Fabric Usage (Y)

XY X²
Licences (X) (in m²)

220 2000 440000 48400

290 2300 667000 84100

240 2200 528000 57600

360 2800 1008000 129600

Sum = 2,643,000 Sum = 319,700

Solve
 Past Question – May 2024
NoMoo Bags specialises in crafting artisanal, handmade bags using vegan leather, and
these creations are available for purchase through their online platform. Each year, the
company offers two distinct seasonal collections: one for Spring/Summer and another for
Autumn/Winter.
NoMoo has been operational for two years. Historically, NoMoo has followed a
make-to-order production model, resulting in extended delivery times for their
customers. NoMoo’s management would like to forecast demand for the upcoming year
(i.e. their third year of operation) with the goal of expediting delivery times.
NoMoo has accumulated the following sales data for analysis in its first two years of
operation.
 Past Question – May 2024

1) Assuming the sales trend from the last two years will be carried into the third year,
estimate the demand for the Spring/Summer and Autumn/Winter collections for the
third year using linear regression analysis with seasonality indexes.
 Solve

t (period) y (sales) ty t2 y2

1 241 241 1 58081

2 189 378 4 35721

3 336 1008 9 112896

4 245 980 16 60025

Sum 1011 2607 30 266723

Solve
 Solve
Using the equation , forecasted values :

t Forecasted Values of y Actual Values of y Actual/Forecast

1 228.90 241 241/228.90= 1.05

2 244.80 189 0.77

3 260.70 336 1.29

4 276.60 245 0.89

 Solve
Seasonality Index for Spring/Summer = (1.05 + 1.29)/2 = 1.17.
Seasonality Index for Autumn/Winter = (0.77 + 0.89)/2 = 0.83.
So, forecasts for next year are:
Spring/Summer = (15.9 × 5 + 213) × 1.17 = 342 bags.
Autumn/Winter = (15.9 × 6 + 213) × 0.83 = 256 bags.