You are on page 1of 16

COMM 225 Excel-based Assignment

Michael McCarthy

ID: 40192118
The two forecasting methods I have chosen are a linear trend equation and a four-
period moving average.
I will compare the two methods by their MAPE (Mean Absolute Percent Error).
Whichever has the lowest MAPE will be the one deemed most accurate.

MAPE formula is as follows: ((sum(|e|/A))/n)*100


|e|= Absolute value of error (ABS function in excel)
A= Actual Demand
n= number of periods

e= A-F (in excel you simply click on the cells to put them in the formula)

Average for previous four periods=(D1+D2+D3+D4)/4 (in excel, AVERAGE function)

The demand data given in the excel spreadsheet is shown on the next page.
Year Month Demand
5637.71
2021 1 8
6833.73
2 1
5390.73
3 1
6514.06
4 7
5 5511.27
5793.77
6 4
6795.09
7 8
6080.08
8 6
6808.38
9 6
6041.80
10 6
5283.31
11 1
6382.96
12 9
6760.45
2022 1 5
2 5820.27
6943.50
3 1
6058.91
4 6
6272.46
5 4
5123.00
6 8
7 5248.18
5575.12
8 4
6024.19
9 3
5006.80
10 2
6673.87
11 1
6830.12
12 5
5303.01
2023 1 3
2 5314.34
5811.86
3 2
5151.26
4 7
5 7432.98
6006.54
6 1
7204.64
7 3

*Numbers have more decimals hidden from this document that were used in the
calculations.

1. Four-Period Moving Forecast


This forecast is calculated by taking the average values of the four periods
preceding the one we’re forecasting; therefore, our forecast only starts at period
five.

Year Month Demand Forecast Error |e| |e|/A


5637.71
2021 1 8
6833.73
2 1
5390.73
3 1
6514.06
4 7
6094.06 - 582.791 0.10574
5 5511.27 2 582.792 8 5
5793.77 - 268.676 0.04637
6 4 6062.45 268.676 1 3
6795.09 992.637 992.637 0.14608
7 8 5802.46 9 9 1
6080.08 6153.55 73.4660 0.01208
8 6 2 -73.466 3 3
9 6808.38 6045.05 763.328 763.328 0.11211
6 7 6 6 6
6041.80 6369.33 327.530 0.05421
10 6 6 -327.53 5 1
5283.31 6431.34 - 1148.03 0.21729
11 1 4 1148.03 3 4
6382.96 6053.39 0.05163
12 9 7 329.572 329.572 3
6760.45 6129.11 0.09338
2022 1 5 8 631.337 631.337 7
6117.13 - 296.865 0.05100
2 5820.27 5 296.866 6 5
6943.50 6061.75 881.749 881.749 0.12698
3 1 1 3 3 9
6058.91 6476.79 - 417.882
4 6 9 417.883 8 0.06897
6272.46 6395.78 - 123.321 0.01966
5 4 5 123.322 5 1
5123.00 6273.78 - 1150.77
6 8 7 1150.78 9 0.22463
6099.47 - 851.292 0.16220
7 5248.18 2 851.292 4 7
5575.12 5675.64 - 100.517
8 4 2 100.518 7 0.01803
6024.19 5554.69 469.499 469.499 0.07793
9 3 4 1 1 6
5006.80 5492.62 - 485.824 0.09703
10 2 6 485.824 1 3
6673.87 5463.57 1210.29 1210.29 0.18134
11 1 5 6 6 8
6830.12 5819.99 1010.12 1010.12 0.14789
12 5 8 7 7 3
5303.01 6133.74 - 830.735 0.15665
2023 1 3 8 830.735 2 3
5953.45 - 639.112 0.12026
2 5314.34 3 639.113 5 2
5811.86 6030.33 - 218.475 0.03759
3 2 7 218.476 5 1
5151.26 5814.83 - 663.567 0.12881
4 7 5 663.568 9 6
0.27416
5 7432.98 5395.12 2037.86 2037.86 5
6006.54 5927.61 78.9288 78.9288
6 1 2 9 9 0.01314
7204.64 6100.66 1103.98 1103.98 0.15323
7 3 2 1 1 2
From these results we calculated the MAPE using an n of 27 and the sum of all (|
e|/A)= 2.898485. The resulting MAPE= 10.74%

Here’s a graph comparing the Actual vs our forecast:

As you can see, it generally follows along the right direction (up or down) of the
Actual demand.

Here’s a graph of what the forecast will look like until December 2024 if we
choose this method:
2. Linear Trend Equation

To find the linear trend equation for our data, we simply highlight all the data
points in excel and insert them into a chart. We chose a scatter plot to enable an
easier viewing of the trendline. Once the chart is made, we can right-click on the
data points and find the linear trendline.

We can also automatically show the forecast for the demand until December
2024.
The trendline shows a slightly downward sloping demand curve and yields the
equation of y=-0.0316x+6053.2

The X values start at 1 and are calculated up to x=48, as we have 31 data points
and we need to forecast the next 17 periods (December 2024).

We must now plug the x values 1-31 into the equation and find the errors
between our forecasted y values and the actual demand values.

Linear
Period Forecast Error |e| |e|/A
415.450 0.07369
1 6053.1684 -415.45 2 1
780.594 780.594 0.11422
2 6053.1368 7 7 7
- 662.373 0.12287
3 6053.1052 662.374 9 3
460.993 460.993 0.07076
4 6053.0736 1 1 9
- 541.771 0.09830
5 6053.042 541.772 9 3
- 259.236 0.04474
6 6053.0104 259.237 6 4
742.119 742.119 0.10921
7 6052.9788 5 5 4
0.00446
8 6052.9472 27.139 27.139 4
755.470 755.470 0.11096
9 6052.9156 1 1 2
- 11.0784 0.00183
10 6052.884 11.0785 8 4
- 769.541 0.14565
11 6052.8524 769.541 1 5
330.148 330.148 0.05172
12 6052.8208 3 3 3
707.665 707.665 0.10467
13 6052.7892 7 7 7
- 0.03994
14 6052.7576 232.488 232.488 5
890.774 890.774 0.12828
15 6052.726 6 6 9
6.22131 6.22131 0.00102
16 6052.6944 4 4 7
219.800 219.800 0.03504
17 6052.6628 9 9 2
- 929.623
18 6052.6312 929.623 1 0.18146
0.15327
19 6052.5996 -804.42 804.42 6
- 477.443 0.08563
20 6052.568 477.444 9 8
- 28.3433 0.00470
21 6052.5364 28.3434 8 5
1045.70 0.20885
22 6052.5048 -1045.7 3 6
621.397 621.397 0.09310
23 6052.4732 9 9 9
777.683 777.683 0.11386
24 6052.4416 1 1 1
- 749.397 0.14131
25 6052.41 749.397 5 5
- 738.038 0.13887
26 6052.3784 738.038 3 7
- 240.485 0.04137
27 6052.3468 240.485 2 8
- 901.048 0.17491
28 6052.3152 901.048 4 8
1380.69 1380.69 0.18575
29 6052.2836 7 7 3
- 45.7109
30 6052.252 45.7109 3 0.00761
1152.42 1152.42 0.15995
31 6052.2204 3 3 6
Total 2.94815

The MAPE using a n value of 31 is 9.51016%

The linear trend method has a lower MAPE compared to the moving average
method. The linear trend is the method we will use to forecast the demand
values.

The forecasted demand values from August 2023 (period 32) until December 2024
(period 48) are calculated by inserting the period values into the linear equation.
Period Demand
6052.18
32 9
6052.15
33 7
6052.12
34 6
6052.09
35 4
6052.06
36 2
6052.03
37 1
6051.99
38 9
6051.96
39 8
6051.93
40 6
6051.90
41 4
6051.87
42 3
6051.84
43 1
44 6051.81
6051.77
45 8
6051.74
46 6
6051.71
47 5
6051.68
48 3

The sum of all these values equals the total forecasted demand, which is
102,882.9, rounded we get 102,883 tons of resin. The sum for just 2024 (=sum(37
to 48)) is 72,622.28, rounds to 72,622 tons.
We can now compute the inventory management section.

Inventory Management

- The resin costs $50 per ton.


- Finn uses 80,000 tons per year.
- Ordering costs total $500.
- Holding costs are 1% of the product value each month.
- Supplier 1 delivers the amount all in one shipment.
- Supplier 2 can deliver one shipment per day, covering two days of demand.
- Capacity limit at the warehouse is 1500 tons at a time.
- Finn operates 365 days a year.
- Forecasted demand is 72,622 tons this upcoming year.

The formula used to calculate the Economic Order Quantity (from the class slides)

2DS 2(Annual Demand) (Order or Setup Cost)


QO = =
H Annual Holding Cost
Q D
TC  H  S
2 Q

We are told to use the updated (forecasted) demand, which is 72,622 tons (D)
Ordering costs are $500 (S)
Holding costs per unit are calculated in excel by (0.01)*(12)*50= $6 (H)

Supplier 1:

Plugging our values into the formula, we get a EOQ of 3479.03 units (3479).
This value does not fit in the warehouse (1500 max) if we receive the full order at
once, so the new EOQ is 1500 units.

The number of orders in the year will be D/EOQ= 48.41, rounded up to 49 orders
(7.44=8 days in between orders.)

No lead time is stated, which implies instant delivery (lead time= 0), so our
demand during lead time is D/365= 198.96. Let’s assume a safety stock of 1.5x the
daily demand to account for fluctuations. 198.96*1.5= ROP = 298.45 =299 tons.

Once inventory reaches 299 tons, we should reorder from Supplier 1.

Total cost is calculated using the formula above and yields a cost of $24,207.33

Supplier 2:

The frequency of these deliveries is not stated so we must assume they are every
day subsequently until the order amount is achieved. Potential negotiations can
change this, resulting in a more efficient inventory management.
Daily demand is 199 tons, so two-day demand is 398 tons. If we try to reach the
best EOQ of 3479, we do 3479/398 = 8.74 = 9 days to fulfil order. The 9-day
demand (inventory used) is 199*9= 1791 – 3479= |1688|, which means that this
EOQ is not achievable with the capacity constraint.

We can achieve an EOQ of 2786 tons (found by doing 398*n (we got 7 days) until
the following subtraction yields a value less than 1500: (199*7) =1393-2786= |
1393|.

With a lead time of 2 days and a safety stock of 1.5x daily demand, we get:
ROP= (2*198.96)*1.5= 596.88= 597 tons.

Once inventory drops to 597 tons remaining, a new order should be placed.

Total Cost is calculated using the formula listed previously, and is (8358+
13,033.38)= $21,391.38

Supplier 2 has a lower Total cost than Supplier 1.


Recommendations

We would recommend choosing supplier 2, assuming the forecast method we


chose (linear trend) is the one you proceed with. Supplier 2 has a lower total cost
than Supplier 1 for the year of 2024, assuming no further negotiations or
discounts.

To potentially improve the forecast, you could perform an exponential smoothing


analysis, as well as others, and compare them using various measures of error
(Mean Absolute Deviation is an example). Between the 4-period moving average
and the linear trend equation, the linear trend equation had a lower MAPE,
meaning it is a more accurate forecast.

Gathering more data such as the production rate for the products, can enhance
the analysis of all phases of production instead of just two.

Overall, we recommend using a linear trend analysis to forecast your demand, and
purchasing resin from Supplier 2, as their flexibility allows you to achieve a higher
EOQ. Look into potentially negotiating even better terms with both suppliers to
lower total cost further.

You might also like