FORECAST PROPOSAL FOR INTEK TAPES

Project work submitted to:

University at Buffalo, The State University at New York

(in partial fulfillment of the requirements for the credits in
IE505 Production Planning and Control)

By:
Group #12
Nilesh Ananthanarayanan (50092066)
Krishnaraj Muthukumar (50097868)
Santosh Kumar Nandakumar (50095660)
Nithin Punaroor Narayanan (50097880)
Arvind Korkadu Sucharitha Sridhar (50095581)
1

OBJECTIVE
The objective of this project is to propose a accurate
forecast model for Intek Tapes Private Ltd.

COMPANY DESCRIPTION
Intek Tapes Private Limited is a company that is
involved in the manufacture of a wide range of
pressure sensitive adhesive tapes.
These tapes are marketed under the renowned
Fixon brand name.
Promoted in 1988; 18,000 sqft. Manufacturing plant.
Products that are designed to suit an array of
specifications in the Leather, Electrical, Electronics,
Printing and Packaging Industry.
Electrical, Shoe upper, Leather goods, Double sided,
Industrial, Masking and Foil tapes.
3

PRODUCT DETAILS
Product
F-620
F-620 YELLOW
F-389

Unit Cost (in INR)

34.81
73.43
1685.40

F-317
F-391
F-372

223.20
458.30
164.51

F-KP1
F-376

1245
310.42
4

PROBLEM FORMULATION
Currently the company doesnt use any forecast
model
The company calculates its demand by assuming a
growth of 20%
This method is not accurate and the error between
demand and forecast is huge.
This method doesnt take into account factors such
as trend , seasonality.

ANALYSIS
We analyzed the data for trend and seasonality.
Depending on our findings the following methods
could be used.
Forecasting method

Applicability

Moving average

No trend or seasonality

Simple exponential
smoothing

No trend or seasonality

Holts model

Trend but no seasonality

Winters model

Trend and seasonality

LINEAR REGRESSION
Linear regression is an approach to model the
relationship between a scalar dependent variable y
and one or more explanatory variables denoted X.
This method is used to calculate the initial Level
and Trend values.
The excel solver is used to calculate the regression
models

Regression Statistics
Multiple R
0.988621
R Square
0.977372
Adjusted R Square 0.971715
Standard Error
16.93607
Observations
6
ANOVA
df
Regression
Residual
Total

Intercept
X Variable 1

Significan
SS
MS
F
ce F
1 49555.8 49555.8 172.7704 0.000193
4 1147.321 286.8304
5 50703.13

Coefficien Standard
Lower
Upper
ts
Error
t Stat
P-value
95%
95%
3507.321 23.31548 150.4288 1.17E-08 3442.587 3572.056
53.21429 4.048494 13.14422 0.000193 41.97386 64.45471

Lower
95.0%
3442.587
41.97386

Upper
95.0%
3572.056
64.45471

WINTERS METHOD
Exponential smoothing is a technique used to
smooth and forecast a time series without the
necessity of fitting a parametric model.
It is based on a recursive computing scheme , where
the forecasts are updated for each new observation.
Winters method , also referred to as double
exponential smoothing is an extension of
exponential smoothing designed for trended and
seasonal time series.

SMOOTHING CONSTANTS
is a smoothing constant for the level. The current
estimate of the level is a weighted average of all
the past observations , with recent observations
weighted higher then older observations. A higher
value of corresponds to a forecast that is more
responsive to recent observations.
is the smoothing constant for trend. Like a higher
value for corresponds to a forecast that is more
responsive to recent observations, also
corresponds to forecast that is more responsive to
recent observations.
is a smoothing constant for the seasonal factor. 10

SIMPLE EXPONENTIAL

Ft+1 = Ft+2 == Ft+n = Lt

Lt+1 = aDt+1 + (1-a)Lt
L= Level
F= forecast
= Smoothing constant

11

HOLTS METHOD

Ft+1 = Lt + Tt
Ft+n = Lt + nTt
Lt+1 = aDt+1 + (1-a)(Lt + Tt)
Tt+1 = b(Lt+1 - Lt) + (1-b)Tt
F = forecast
L = level
T= trend
= smoothing constant for level
= smoothing constant for trend
12

WINTERS METHOD

Ft+1 = (Lt+Tt)(St+1) and Ft+n = (Lt + nTt)St+n

Lt+1 = a(Dt+1/St+1) + (1-a)(Lt+Tt)
Tt+1 = b(Lt+1 - Lt) + (1-b)Tt
St+p+1 = g(Dt+1/Lt+1) + (1-g)St+1
= smoothing constant for level
= smoothing constant for trend
= smoothing constant for seasonal factor
L= level
T= Trend
S= Seasonal Factor

13

Calculations using Forecasting methods

(using Exponential Smoothing for F-620)
Period

Demand Level Forecast

3810.0

Error

% Error

MAPE

3700

3799.0

3810.0

110.0

2.97

2.97

3450

3764.1

3799.0

349.0

10.12

6.55

3800

3767.7

3764.1

35.9

0.94

4.68

3650

3755.9

3767.7

117.7

3.22

4.31

3900

3770.3

3755.9

144.1

3.69

4.19

3500

3743.3

3770.3

270.3

7.72

4.78

4250

3794.0

3743.3

506.7

11.92

5.8

3850

3799.6

3794.0

56.0

1.46

5.26

3800

3799.6

3799.6

0.4

0.01

4.67

10

4200

3839.7

3799.6

400.4

9.53

5.16
4.837
14

Calculations using Forecasting methods

(using Holts method for F-620)
Period

1
2
3
4
5
6
7
8
9
10

Demand Level

3700
3450
3800
3650
3900
3500
4250
3850
3800
4200

3523.33
3
3575.45
3627.57
3679.69
3731.81
3783.93
3836.05
3888.17
3940.29
3992.41
4044.53

Trend

Forecast

Absolute
Error

PE

MAPE

52.1212
1
52.12
52.12
52.12
52.12
52.12
52.12
52.12
52.12
52.12
52.12

3575.45
3627.57
3679.69
3731.81
3783.93
3836.05
3888.17
3940.29
3992.41
4044.53

124.55
177.57
120.31
81.81
116.07
336.05
361.83
90.29
192.41
155.47

3.37
5.15
3.17
2.24
2.98
9.6
8.51
2.35
5.06
3.7

4.71
4.75
4.64
4.48
4.38
4.69
4.9
4.77
4.78
4.73
4.683
15

Calculations using Forecasting methods

(using Winters model for F-620)
Forecast

D/F ratio

Seasonal
factor

3675
3706.25

3560.536
3613.75
3666.964
3720.179

1.039
0.955
1.036
0.981

1.009
0.97
1.066
0.98

3900

3768.75

3773.393

1.034

3500

3850

3826.607

0.915

7
8
9
10

4250
3850
3800
4200

3862.5
3937.5

3879.821
3933.036
3986.25
4039.464

1.095
0.979
0.953
1.04

Period

Demand

1
2
3
4

3700
3450
3800
3650

Deseasonalized
Demand

16

Calculations using Forecasting methods

(using Winters model for F-620)
Period

Demand

Level

Trend

3507.321

53.21429

Seasonility

Forecast

AbsoluteEr
ror

% Error

MAPE

3700

3571.182

54.279

1.009

3592.581

107.419

2.903216

2.903216

3450

3618.585

53.591

0.97

3516.697

66.697

1.933246

2.418231

3800

3661.431

52.517

1.066

3914.54

114.54

3.014211

2.616891

3650

3715.002

52.622

0.98

3639.669

10.331

0.283041

2.033429

3900

3776.237

53.483

1.012

3812.835

87.165

2.235

2.073743

3500

3808.318

51.343

0.968

3707.169

207.169

5.919114

2.714638

4250

3873.507

52.728

1.063

4102.82

147.18

3.463059

2.821555

3850

3926.469

52.751

0.98

3847.71

2.29

0.059481

2.476296

3800

3956.051

50.434

1.014

4034.929

234.929

6.182342

2.888079

10

4200

4041.974

53.983

0.963

3858.245

341.755

8.137024

3.412973
2.635905
17

MAPE Calculations for different

products
PRODUCT

SIMPLE
EXPONENTIAL
4.837

HOLTS

WINTERS

4.373

2.6359

F-620 yellow

18.095

18.62

16.6769

F-389

29.117

30.539

27.70925

F-317

45.816

46.387

42.265

F-391

29.38

35.725

25.960

F-372

49.326

22.848

27.51834

F-KP1

82.047

80.727

46.71

F-376

57.976

17.811

15.511

F-620

18

Calculations using Forecasting methods

(Savings in cost using Winters model for F-620)
Period
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16

Companys model
3700
3450
3800
3650
3900
3500
4250
3850
3800
4200
5100
4620
4560
5040
6120
5544

Our model
3592.581
3516.697
3914.54
3639.669
3812.835
3707.169
4102.82
3847.71
4034.929
3858.245
4366.29
3625.396
3335.224
2833.283
2435.771
1917.743
Total Error
Unit Cost
Loss

Absolute Error
108
67
115
11
88
208
148
3
235
342
734
995
1225
2207
3685
3627
13798
34.81
480,308.38
19

7000
6000
5000
4000

Demand
Forecast

3000
2000
1000
Demand

0
1

10

11

12

13

14

15

16

20

RESULTS
From the MAPE calculations you can see that winters
model is the most accurate.
If this model is used to forecast demand the
company can save a lot of money
The company can also reduce over head costs such
as inventory and transportation costs by applying
this model.

21

Thank You!
Questions?

22

Happy Thanksgiving!!!

23