Naïve Approach Moving Averages Exponential Smoothing Trend ...
Transcript of Naïve Approach Moving Averages Exponential Smoothing Trend ...
![Page 1: Naïve Approach Moving Averages Exponential Smoothing Trend ...](https://reader031.fdocuments.us/reader031/viewer/2022012018/615c9c431fa52e1f1a31343f/html5/thumbnails/1.jpg)
Quantitative Forecasting MethodsQuantitative Forecasting Methods
• Naïve Approach• Moving Averages• Exponential Smoothing• Trend projection• Linear Regression
• Naïve Approach• Moving Averages• Exponential Smoothing• Trend projection• Linear Regression
![Page 2: Naïve Approach Moving Averages Exponential Smoothing Trend ...](https://reader031.fdocuments.us/reader031/viewer/2022012018/615c9c431fa52e1f1a31343f/html5/thumbnails/2.jpg)
Time Series ModelsTime Series Models
• Future is the function of past• Use series of past data to forecast• Future is the function of past• Use series of past data to forecast
![Page 3: Naïve Approach Moving Averages Exponential Smoothing Trend ...](https://reader031.fdocuments.us/reader031/viewer/2022012018/615c9c431fa52e1f1a31343f/html5/thumbnails/3.jpg)
Associative ModelsAssociative Models
• Causal Models• Involves variables that might influence the
quantity being forecast
• Causal Models• Involves variables that might influence the
quantity being forecast
![Page 4: Naïve Approach Moving Averages Exponential Smoothing Trend ...](https://reader031.fdocuments.us/reader031/viewer/2022012018/615c9c431fa52e1f1a31343f/html5/thumbnails/4.jpg)
Decomposition of Time SeriesDecomposition of Time Series
• Trend- Gradual upward or downward movementof data
• Seasonality-Data that repeats itself after a periodof days
• Cycles- Data pattern that occurs after several years• Random variations- No observable pattern
• Trend- Gradual upward or downward movementof data
• Seasonality-Data that repeats itself after a periodof days
• Cycles- Data pattern that occurs after several years• Random variations- No observable pattern
![Page 5: Naïve Approach Moving Averages Exponential Smoothing Trend ...](https://reader031.fdocuments.us/reader031/viewer/2022012018/615c9c431fa52e1f1a31343f/html5/thumbnails/5.jpg)
Increasing Trend in DataIncreasing Trend in Data
![Page 6: Naïve Approach Moving Averages Exponential Smoothing Trend ...](https://reader031.fdocuments.us/reader031/viewer/2022012018/615c9c431fa52e1f1a31343f/html5/thumbnails/6.jpg)
![Page 7: Naïve Approach Moving Averages Exponential Smoothing Trend ...](https://reader031.fdocuments.us/reader031/viewer/2022012018/615c9c431fa52e1f1a31343f/html5/thumbnails/7.jpg)
Seasonal DataSeasonal Data
![Page 8: Naïve Approach Moving Averages Exponential Smoothing Trend ...](https://reader031.fdocuments.us/reader031/viewer/2022012018/615c9c431fa52e1f1a31343f/html5/thumbnails/8.jpg)
Cyclic DataCyclic Data
![Page 9: Naïve Approach Moving Averages Exponential Smoothing Trend ...](https://reader031.fdocuments.us/reader031/viewer/2022012018/615c9c431fa52e1f1a31343f/html5/thumbnails/9.jpg)
Group• http://groups.yahoo.com/neo/groups/OMS20
13/info
• Term Project
Group• http://groups.yahoo.com/neo/groups/OMS20
13/info
• Term Project
![Page 10: Naïve Approach Moving Averages Exponential Smoothing Trend ...](https://reader031.fdocuments.us/reader031/viewer/2022012018/615c9c431fa52e1f1a31343f/html5/thumbnails/10.jpg)
• Form of weighted moving average– Weights decline exponentially– Most recent data weighted most
• Requires smoothing constant ()– Ranges from 0 to 1– Subjectively chosen
• Involves little record keeping of past data
Exponential Smoothing Method
• Form of weighted moving average– Weights decline exponentially– Most recent data weighted most
• Requires smoothing constant ()– Ranges from 0 to 1– Subjectively chosen
• Involves little record keeping of past data
![Page 11: Naïve Approach Moving Averages Exponential Smoothing Trend ...](https://reader031.fdocuments.us/reader031/viewer/2022012018/615c9c431fa52e1f1a31343f/html5/thumbnails/11.jpg)
Exponential SmoothingNew forecast =previous forecast + (previous actual - previous)
or:where
Ft = Ft-1 + (At-1 - Ft-1)
New forecast =previous forecast + (previous actual - previous)
or:where
Ft = Ft-1 + (At-1 - Ft-1)
Ft-1 = previous forecast
= smoothing constant
Ft = new forecast
At-1 = previous period actual
![Page 12: Naïve Approach Moving Averages Exponential Smoothing Trend ...](https://reader031.fdocuments.us/reader031/viewer/2022012018/615c9c431fa52e1f1a31343f/html5/thumbnails/12.jpg)
• Ft = At - 1 + (1-)At - 2 + (1- )2·At - 3
+ (1- )3At - 4 + ... + (1- )t-1·A0
– Ft = Forecast value– At = Actual value– = Smoothing constant
Exponential Smoothing Equations
• Ft = At - 1 + (1-)At - 2 + (1- )2·At - 3
+ (1- )3At - 4 + ... + (1- )t-1·A0
– Ft = Forecast value– At = Actual value– = Smoothing constant
![Page 13: Naïve Approach Moving Averages Exponential Smoothing Trend ...](https://reader031.fdocuments.us/reader031/viewer/2022012018/615c9c431fa52e1f1a31343f/html5/thumbnails/13.jpg)
Ft = At - 1 + (1- ) At - 2 + (1- )2At - 3 + ...
Forecast Effects ofSmoothing Constant
Weights
Prior Period
2 periods ago
(1 - )
3 periods ago
(1 - )2
= Prior Period
2 periods ago
(1 - )
3 periods ago
(1 - )2
=
= 0.10
= 0.90
10% 9% 8.1%
90% 9% 0.9%
![Page 14: Naïve Approach Moving Averages Exponential Smoothing Trend ...](https://reader031.fdocuments.us/reader031/viewer/2022012018/615c9c431fa52e1f1a31343f/html5/thumbnails/14.jpg)
Table 5.4
![Page 15: Naïve Approach Moving Averages Exponential Smoothing Trend ...](https://reader031.fdocuments.us/reader031/viewer/2022012018/615c9c431fa52e1f1a31343f/html5/thumbnails/15.jpg)
Exponential Smoothing with Trend Adjustment
• Simple exponential smoothing - first-ordersmoothing
• Trend adjusted smoothing - second-ordersmoothing
• Low gives less weight to more recenttrends, while high gives higher weight tomore recent trends
• Simple exponential smoothing - first-ordersmoothing
• Trend adjusted smoothing - second-ordersmoothing
• Low gives less weight to more recenttrends, while high gives higher weight tomore recent trends
![Page 16: Naïve Approach Moving Averages Exponential Smoothing Trend ...](https://reader031.fdocuments.us/reader031/viewer/2022012018/615c9c431fa52e1f1a31343f/html5/thumbnails/16.jpg)
Selecting the Smoothing Constant ()
Select to minimize:Mean Absolute Deviation = MAD
n
|errorsforecast|Σ
Mean Square Error = MSEn
errors)Σ(forecast 2
Mean Square Error = MSEn
errors)Σ(forecast 2
Mean Absolute Percent Error = MAPE
actual
errorforecastΣ
n
1
Bias =forecast errors
![Page 17: Naïve Approach Moving Averages Exponential Smoothing Trend ...](https://reader031.fdocuments.us/reader031/viewer/2022012018/615c9c431fa52e1f1a31343f/html5/thumbnails/17.jpg)
Error Analysis (Alpha=0.1)
![Page 18: Naïve Approach Moving Averages Exponential Smoothing Trend ...](https://reader031.fdocuments.us/reader031/viewer/2022012018/615c9c431fa52e1f1a31343f/html5/thumbnails/18.jpg)
Error Analysis (Alpha=0.5)
![Page 19: Naïve Approach Moving Averages Exponential Smoothing Trend ...](https://reader031.fdocuments.us/reader031/viewer/2022012018/615c9c431fa52e1f1a31343f/html5/thumbnails/19.jpg)
Exponential Smoothing with Trend Adjustment
• Simple exponential smoothing - first-ordersmoothing
• Trend adjusted smoothing - second-ordersmoothing
• Low gives less weight to more recenttrends, while high gives higher weight tomore recent trends
• Simple exponential smoothing - first-ordersmoothing
• Trend adjusted smoothing - second-ordersmoothing
• Low gives less weight to more recenttrends, while high gives higher weight tomore recent trends
![Page 20: Naïve Approach Moving Averages Exponential Smoothing Trend ...](https://reader031.fdocuments.us/reader031/viewer/2022012018/615c9c431fa52e1f1a31343f/html5/thumbnails/20.jpg)
Exponential Smoothing withTrend Adjustment
Forecast including trend (FITt+1) =new forecast (Ft) + trend correction(Tt)
whereTt = (1 - )Tt-1 + (Ft – Ft-1)
where
Tt = smoothed trend for period t
Tt-1 = smoothed trend for the preceding period
= trend smoothing constant
Ft = simple exponential smoothed forecast for period t
Ft-1 = forecast for period t-1
![Page 21: Naïve Approach Moving Averages Exponential Smoothing Trend ...](https://reader031.fdocuments.us/reader031/viewer/2022012018/615c9c431fa52e1f1a31343f/html5/thumbnails/21.jpg)
Exponential Forecast with TREND(Alpha=0.1, Beta=0.5)
![Page 22: Naïve Approach Moving Averages Exponential Smoothing Trend ...](https://reader031.fdocuments.us/reader031/viewer/2022012018/615c9c431fa52e1f1a31343f/html5/thumbnails/22.jpg)
Trend Projection
General regression equation:
22X
nXY-XYb
interceptaxis-Ya
variable)(dependent
predictedbe
to variable theof
valuecomputedY
where
bXaY
nX
22X
nXY-XYb
interceptaxis-Ya
variable)(dependent
predictedbe
to variable theof
valuecomputedY
where
bXaY
nX
![Page 23: Naïve Approach Moving Averages Exponential Smoothing Trend ...](https://reader031.fdocuments.us/reader031/viewer/2022012018/615c9c431fa52e1f1a31343f/html5/thumbnails/23.jpg)
Trend Projections
![Page 24: Naïve Approach Moving Averages Exponential Smoothing Trend ...](https://reader031.fdocuments.us/reader031/viewer/2022012018/615c9c431fa52e1f1a31343f/html5/thumbnails/24.jpg)
45
51
54
61
66
70
74
78
85
89
40
50
60
70
80
90
100
100 110 120 130 140 150 160 170 180 190Temp (Deg C)
Yiel
d (%
)
Actual Data Predited Values
Graph: Actual vs Fitted Line
Trend Line
45
51
54
61
66
70
74
78
85
89
40
50
60
70
80
90
100
100 110 120 130 140 150 160 170 180 190Temp (Deg C)
Yiel
d (%
)
Actual Data Predited Values
Actual demand line
![Page 25: Naïve Approach Moving Averages Exponential Smoothing Trend ...](https://reader031.fdocuments.us/reader031/viewer/2022012018/615c9c431fa52e1f1a31343f/html5/thumbnails/25.jpg)
Seasonal Variations
Month SalesDemand
AverageTwo-YearDemand
AverageMonthlyDemand
SeasonalIndex
Year1
Year2
80 100 90 94 0.957
75 85 80 94 0.85180 90 85 94 0.904
Jan
Feb
Mar 80 90 85 94 0.904
90 110 100 94 1.064
Mar
Apr
May 115 131 123 94 1.309… … … … … …
Total of Average Demand = 1,128Seasonal Index:
= Average 2 -year demand/Average monthly demand
1128/12
![Page 26: Naïve Approach Moving Averages Exponential Smoothing Trend ...](https://reader031.fdocuments.us/reader031/viewer/2022012018/615c9c431fa52e1f1a31343f/html5/thumbnails/26.jpg)
If total demand/year for year 3=1200, thenmonthly forecast= (1200/12)* seasonal index
Month SalesDemand
AverageTwo-YearDemand
AverageMonthlyDemand
SeasonalIndex
Year1
Year2
80 100 90 94 0.957
75 85 80 94 0.85180 90 85 94 0.904
Jan
Feb
Mar
95.785.1
90.480 90 85 94 0.904
90 110 100 94 1.064
Mar
Apr
May 115 131 123 94 1.309… … … … … …
Total of Average Demand = 1,128Seasonal Index:
= Average 2 -year demand/Average monthlydemand
1128/12
90.4106.4130.9