Case : Altavox Electronics

Demand Management and Forecasting

Presented by- Anubhav Sood 12PGP009
Bhairav Mehta 12PGP013
Jamal Shahid 12PGP019
S. Sreejith 12PGP037
Vivek Kerketta 12PGP053
Anshul Tripathi 12FPM001

Case Study Premise
Altavox is a manufacturer and distributor of many different electronic
instruments and devices, including digital/ analog multimeters etc.
Altavox sells a line of test meters (Model VC202) through six distributors to
retail stores in United States.
Distributors : Atlanta, Boston, Chicago, Dallas and Los Angeles.
VC202 has been steady seller over the years. Altavox does not consider this as
a Seasonal Product but there is variability in demand.
Management wants us to experiment with forecasting models to be used in a
new system to be implemented.
Two models : Simple moving average and Exponential Smoothing.
Q.1 - Q.1)Consider using a simple average model. Experiment with models
using five weeks and three weeks past data. The past data in each region
is given . Evaluate the forecast that would have been made over the past
13 weeks using the MAD and tracking signal as criteria
Week` -5 -4 -3 -2 -1
Atlanta 45 38 30 58 37
Boston 62 18 48 40 35
Chicago 62 22 72 44 48
Dallas 42 35 40 64 43
LA 43 40 54 46 35
Total 254 153 244 252 198
The Evaluated Forecasts that would have been made
over the past 13 weeks using mean absolute deviation
and tracking signal as criteria.
Based on the 3 period moving averages, the following data is evaluated
MAD = 61.87 (Mean Absolute Deviation)
Average Demand over 13 weeks = 222
MAPE = MAD/Average Demand * 100 = 27.83 (MAPE = Mean Absolute Percent
Error)
Accuracy = 100 MAPE = 72.17
Hence, 3 Period Moving Averages method is 72.17 % accurate in forecasting
the demand given in the problem.

The Evaluated Forecasts that would have been made
over the past 13 weeks using mean absolute deviation
and tracking signal as criteria.
Based on the 5 period moving averages, the following data is evaluated
MAD = 62.20 (Mean Absolute Deviation)
Average Demand over 13 weeks = 222
MAPE = MAD/Average Demand * 100 = 27.98 (MAPE = Mean Absolute Percent Error)
Accuracy = 100 MAPE = 72.02
Hence, 5 Period Moving Averages method is 72.02 % accurate in forecasting the
demand given in the problem.
As the number of periods of moving averages increases, the MAD value goes on
increasing. Thus the forecasted value increasingly deviates from the actual value.
This happens as the forecasted curve becomes flatter as the periods of moving
averages increase.

The Evaluated Forecasts that would have been made
over the past 13 weeks using mean absolute deviation
and tracking signal as criteria.
Tracking signal is a measurement that indicates whether the forecast average
is keeping pace with any genuine upward or downward changes in demand. As
used in forecasting, the tracking signal is the number of mean absolute
deviations that the forecast value is above or below the actual occurrence.
Tracking Signal TS = RSFE/MAD ( RSFE = Running sum of forecast errors i.e
with +/-)
The MAD, RSFE and Tracking signal calculation has been shown in the excel
sheet attached.



Q.2)Next consider using a simple exponential smoothing model. In your analysis
test two alpha values 0.2 and 0.4 .Use the same criteria for evaluating the models
as in part 1.Assume the initial previous forecast for the model as in part 1.Assume
that the initial previous forecast for the model using an alpha value of 0.2 is the
past three week average. For the model using an alpha of 0.4 Assume the previous
forecast is the past five week average.

Based on the Exponential smoothing method and Smoothing constant alpha
given as 0.2 with initial forecast calculated with 3 period moving average,the
following data is evaluated.
F
t
= 0.2 A
t-1
+ 0.8 (F
t-1
) ( From F
t
= F
t-1
+ 0.2 (A
t-1
- F
t-1
))
MAD = 57.47
Average Demand over 13 weeks = 222
MAPE = MAD/Average Demand * 100 = 25.85 (MAPE = Mean Absolute Percent
Error)
Accuracy = 100 MAPE = 74.15
Hence, Exponential Smoothing method is 74.15 % accurate in forecasting the
demand given in the problem.

Q.2)Next consider using a simple exponential smoothing model. In your analysis
test two alpha values 0.2 and 0.4 .Use the same criteria for evaluating the models
as in part 1.Assume the initial previous forecast for the model as in part 1.Assume
that the initial previous forecast for the model using an alpha value of 0.2 is the
past three week average. For the model using an alpha of 0.4 Assume the previous
forecast is the past five week average.


Based on the Exponential smoothing method and Smoothing constant alpha given
as 0.4 with initial forecast value calculated with 5 period moving average,the
following data is evaluated.
F
t
= 0.4 A
t-1
+ 0.6 (F
t-1
) ( From F
t
= F
t-1
+ 0.4 (A
t-1
- F
t-1
))
MAD = 60.15
Average Demand over 13 weeks = 222
MAPE = MAD/Average Demand * 100 = 27.06 (MAPE = Mean Absolute Percent Error)
Accuracy = 100 MAPE = 72.94
Hence, Exponential Smoothing method is 72.94 % accurate in forecasting the
demand given in the problem. However, Forecast with alpha = 0.2 is more
accurate than alpha = 0.4.
Q 3 )Altavox is considering a new option for distributing the model VC 202 where , instead of
using five vendors, only a single vendor would be used . Evaluate this option by analysing how
accurate the forecast would be based on the demand aggregated across all regions . Use the
model that you think is best from your analysis of part 1 and part 2.Use a new criterion that is
calculated by taking the MAD and dividing by the average demand . This is called MAPE and
gauge the error of forecast as percent of the average demand . What are the advantage ad
disadvantage of aggregating demand from a forecasting view . Are there other things thats
should be considered when going from multiple to single distributor.

Considering that the entire demand is given to a single supplier, we focus on the aggregate demand.
Since we evaluated that Exponential smoothing with alpha = 0.2 is the best method to forecast, we
have used the same to forecast the aggregated values.
Following data was evaluated:
MAD = 30.06
MAPE = 13.52
Accuracy of aggregated model of forecast = 86.48 %
Thus , as per the Golden Rule : Aggregated Demand is always accurate than the disaggregate
demand.
Advantages:
1) Aggregating demand reduces the total error as compared to individual demand.
2) Forecast predication is better in aggregated demand.
Multiple distributor over single distributor :
While going from multiple distributors to single distributors a lot of factors have to be taken in to
consideration like the inventory costs, transport cost, proximity to Wholesalers and Retailers as
well as storage costs.
The storage cost will over all be reduced as the forecast prediction is more accurate in case of
aggregated demand than in case of single distributor
Thank You