Description
This is a PPT describes about demand forecasting methods.
640K ought to be enough for anybody. -- Bill Gates, 1981 Prediction is very difficult, especially it it’s about the future - Nils Bhor I always avoid prophesying beforehand because it is much better to prophesy after the event has already taken place - Winston Churchill An economist is an expert who will know tomorrow why the things he predicted yesterday didn’t happen today - Evan Esar
?
Cisco, which had installed a variety of sophisticated supply chain management software, still found itself in 2001 with an embarrassing $2.25 billion in excess inventory as a result of inflated forecasts of demand. Sony Play station II could meet only 25% of potential demand in the Christmas during year of introduction
?
Customer Order
SOURCE
Customer Order
MAKE COMPONENTS
ASSEMBLY
DELIVERY
MTS
SOURCE
MAKE COMPONENTS
ASSEMBLY
Customer Order
DELIVERY
MTO
SOURCE
MAKE COMPONENTS
ASSEMBLY
DELIVERY
CTO
3.0
2.5
2.0
Before After
1.5
1.0 1 2 3 4 5 6 7 8 9 10 11 12
Peak to Minimum ration down from 3 to 1.3 Average sales is same
? ?
Who should Forecast? What is appropriate time Horizon
? Unit of time ?
?
What is Appropriate Level of Aggregation
Pyramid Forecasting
Total Business
Product Lines
Individual Items
? ? ? ?
?
Short Term Forecasting at more SKU level Little time to react for forecast errors Inventory planning, production scheduling and transportation decisions are based on them. Time Horizon is (one week-three months).
? ? ?
?
Medium Term Time horizon is (three months- one year) For aggregate and workforce planning decisions. Forecasting is at ‘moderate’ aggregate level.
? ?
?
Time horizon is more than two years. Forecasts are at aggregate levels such as sales in tons, liters. Needed for process selection, location decisions.
Uses of Forecasting for Operations Decisions
Capacity planning facilities/ Process Design Aggregate planning Scheduling Inventory Management
Time Horizon
Accuracy Required
Forecasting Method
Long
Medium
Qualitative and causal Causal and time series Time Series Time Series
Medium Short Short
High Highest Highest
•Qualitative • Delphi • Market surveys
• Lifecycles analogy
•Quantitative Forecasting
•Time series Forecasting
•Causal Models
• Moving averages
• Exponential smoothing
• Regression Trend Analysis • Time Series decomposition • Box-Jenkins
Decomposition of Time series data
Seasonal pattern Time
DEMAND
Trend
Level
?
Seasonal pattern
? ? ? ? Quarter effect Month effect Week effect Day of the week effect
Use past data to determine appropriate forecasting model & estimate relevant parameters
? Clean up the data ( promotions, abnormal events) - Draw scatter diagram of past data. Examine data for visual patterns of trend and/or seasonality. Use appropriate model based on kind of forms and patterns observed in the data. Estimate relevant parameters for patterns observed in past data Estimate forecasting error
Demand (t) = ( Level(t) + Trend parameter * t) * seasonality parameter(t) + Random - No Trend/Seasonality
? Moving average ? Exponential smoothing
Ft+1 = ? Dt + (1- ? ) Ft
-
Ft+1 = (Dt + D t-1 +…+ Dt-n+1 )/n
Trend/Seasonality Component Present
- Trend Ft = Level(t) + b t -Trend & Seasonality Ft = (Level(t) + b t) * St
?
Use past data to determine appropriate forecasting model & estimate relevant parameters
? ? ? ? ? Prepare the data Identify seasonal index Deseasonalise data Find trend Estimate forecasting error
?
Preparing forecast for future time period
? Forecast deseasonlised estimate ? Multiply above estimate with appropriate seasonal indices
? ?
?
Impact of local festivals. Impact of promotions and other abnormal events. Sales is not same as demand.
Sales data for a Publishing Company
25000
20000
15000
10000
5000
0 Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec Jan Feb Mar Apr May Jun Jul
Annual Sales in million units 1.2 1 0.8 0.6 0.4 0.2 0 1 4 7 10 13 16 19 22 25 28 31 34 37 40 Annual Sales in million units
m= Total market size p= Innovation parameter ( External Influence) q= Imitation parameters ( Internal influence) S(t) = sales in time period t N(t) = Cumulative sales up to time t S(t) = p ( m – N(t-1)) + q ( N(t-1)/m) (( m – N(t-1))
Bass Model : Impact of parameters
120.00 100.00
Market Penetrat ion
80.00 60.00 40.00 20.00 0.00 1 3 5 7 9 11 13 15 17 19
p=0.01,q=0.1 p=0.1,q=0.01 p=0.1,q=0.1
?
? ? ? ?
User and system sophistication Time and resource available Use or decision characteristic Data availability Data pattern
?
"Prediction is very difficult, especially if it's about the future." --Nils Bohr, Nobel laureate in Physics This quote serves as a warning of the importance of testing a forecasting model out-of-sample. It's often easy to find a model that fits the past data well--perhaps too well!--but quite another matter to find a model that correctly identifies those features of the past data which will be replicated in the future.
?
? ?
The Forecasting is always wrong The Longer the forecast horizon, the worse the forecast
? Short life cycle products : Focus on Sense & respond
Forecasting Errors
Forecast Error
Forecast at SKU Level Forecast at Aggregate Product Level
Time * * Time before the event for which forecast is made
?
? ?
Selecting forecasting model. Monitoring forecasting process: Making optimal decisions about safety stock and safety capacity
Forecast error (t) = demand (t) - forecast (t)
Mean Error (ME) : (? e(t)) / n Mean Absolute Deviation (MAD = (? ? e(t) ? )/n Mean Square Error ( MSE ) = (? e(t)2) / n Mean absolute percentage error( MAPE) = (? ? e(t)/D(t) ? )*100/n
?
? ?
Better Forecasting Methods Build more flexibility into supply chain Reduce Lead-time
?
Involvement of regional sales force in forecasting. Forecast is often confused with goals.
? Can inflate the forecast numbers
?
?
Impact of performance measures.
? Lower forecast numbers.
?
As a SC planner, you would like to forecast the sales for Amazon for the 2006 last quarter. Data for the last few years is as below.
2002
847 806 851 1429
2003
1084 1100 1134 1946
2004
1530 1387 1463 2541
2005
1902 1753 1858 2977
2006
2279 2139 2309
doc_102956488.pptx
This is a PPT describes about demand forecasting methods.
640K ought to be enough for anybody. -- Bill Gates, 1981 Prediction is very difficult, especially it it’s about the future - Nils Bhor I always avoid prophesying beforehand because it is much better to prophesy after the event has already taken place - Winston Churchill An economist is an expert who will know tomorrow why the things he predicted yesterday didn’t happen today - Evan Esar
?
Cisco, which had installed a variety of sophisticated supply chain management software, still found itself in 2001 with an embarrassing $2.25 billion in excess inventory as a result of inflated forecasts of demand. Sony Play station II could meet only 25% of potential demand in the Christmas during year of introduction
?
Customer Order
SOURCE
Customer Order
MAKE COMPONENTS
ASSEMBLY
DELIVERY
MTS
SOURCE
MAKE COMPONENTS
ASSEMBLY
Customer Order
DELIVERY
MTO
SOURCE
MAKE COMPONENTS
ASSEMBLY
DELIVERY
CTO
3.0
2.5
2.0
Before After
1.5
1.0 1 2 3 4 5 6 7 8 9 10 11 12
Peak to Minimum ration down from 3 to 1.3 Average sales is same
? ?
Who should Forecast? What is appropriate time Horizon
? Unit of time ?
?
What is Appropriate Level of Aggregation
Pyramid Forecasting
Total Business
Product Lines
Individual Items
? ? ? ?
?
Short Term Forecasting at more SKU level Little time to react for forecast errors Inventory planning, production scheduling and transportation decisions are based on them. Time Horizon is (one week-three months).
? ? ?
?
Medium Term Time horizon is (three months- one year) For aggregate and workforce planning decisions. Forecasting is at ‘moderate’ aggregate level.
? ?
?
Time horizon is more than two years. Forecasts are at aggregate levels such as sales in tons, liters. Needed for process selection, location decisions.
Uses of Forecasting for Operations Decisions
Capacity planning facilities/ Process Design Aggregate planning Scheduling Inventory Management
Time Horizon
Accuracy Required
Forecasting Method
Long
Medium
Qualitative and causal Causal and time series Time Series Time Series
Medium Short Short
High Highest Highest
•Qualitative • Delphi • Market surveys
• Lifecycles analogy
•Quantitative Forecasting
•Time series Forecasting
•Causal Models
• Moving averages
• Exponential smoothing
• Regression Trend Analysis • Time Series decomposition • Box-Jenkins
Decomposition of Time series data
Seasonal pattern Time
DEMAND
Trend
Level
?
Seasonal pattern
? ? ? ? Quarter effect Month effect Week effect Day of the week effect
Use past data to determine appropriate forecasting model & estimate relevant parameters
? Clean up the data ( promotions, abnormal events) - Draw scatter diagram of past data. Examine data for visual patterns of trend and/or seasonality. Use appropriate model based on kind of forms and patterns observed in the data. Estimate relevant parameters for patterns observed in past data Estimate forecasting error
Demand (t) = ( Level(t) + Trend parameter * t) * seasonality parameter(t) + Random - No Trend/Seasonality
? Moving average ? Exponential smoothing
Ft+1 = ? Dt + (1- ? ) Ft
-
Ft+1 = (Dt + D t-1 +…+ Dt-n+1 )/n
Trend/Seasonality Component Present
- Trend Ft = Level(t) + b t -Trend & Seasonality Ft = (Level(t) + b t) * St
?
Use past data to determine appropriate forecasting model & estimate relevant parameters
? ? ? ? ? Prepare the data Identify seasonal index Deseasonalise data Find trend Estimate forecasting error
?
Preparing forecast for future time period
? Forecast deseasonlised estimate ? Multiply above estimate with appropriate seasonal indices
? ?
?
Impact of local festivals. Impact of promotions and other abnormal events. Sales is not same as demand.
Sales data for a Publishing Company
25000
20000
15000
10000
5000
0 Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec Jan Feb Mar Apr May Jun Jul
Annual Sales in million units 1.2 1 0.8 0.6 0.4 0.2 0 1 4 7 10 13 16 19 22 25 28 31 34 37 40 Annual Sales in million units
m= Total market size p= Innovation parameter ( External Influence) q= Imitation parameters ( Internal influence) S(t) = sales in time period t N(t) = Cumulative sales up to time t S(t) = p ( m – N(t-1)) + q ( N(t-1)/m) (( m – N(t-1))
Bass Model : Impact of parameters
120.00 100.00
Market Penetrat ion
80.00 60.00 40.00 20.00 0.00 1 3 5 7 9 11 13 15 17 19
p=0.01,q=0.1 p=0.1,q=0.01 p=0.1,q=0.1
?
? ? ? ?
User and system sophistication Time and resource available Use or decision characteristic Data availability Data pattern
?
"Prediction is very difficult, especially if it's about the future." --Nils Bohr, Nobel laureate in Physics This quote serves as a warning of the importance of testing a forecasting model out-of-sample. It's often easy to find a model that fits the past data well--perhaps too well!--but quite another matter to find a model that correctly identifies those features of the past data which will be replicated in the future.
?
? ?
The Forecasting is always wrong The Longer the forecast horizon, the worse the forecast
? Short life cycle products : Focus on Sense & respond
Forecasting Errors
Forecast Error
Forecast at SKU Level Forecast at Aggregate Product Level
Time * * Time before the event for which forecast is made
?
? ?
Selecting forecasting model. Monitoring forecasting process: Making optimal decisions about safety stock and safety capacity
Forecast error (t) = demand (t) - forecast (t)
Mean Error (ME) : (? e(t)) / n Mean Absolute Deviation (MAD = (? ? e(t) ? )/n Mean Square Error ( MSE ) = (? e(t)2) / n Mean absolute percentage error( MAPE) = (? ? e(t)/D(t) ? )*100/n
?
? ?
Better Forecasting Methods Build more flexibility into supply chain Reduce Lead-time
?
Involvement of regional sales force in forecasting. Forecast is often confused with goals.
? Can inflate the forecast numbers
?
?
Impact of performance measures.
? Lower forecast numbers.
?
As a SC planner, you would like to forecast the sales for Amazon for the 2006 last quarter. Data for the last few years is as below.
2002
847 806 851 1429
2003
1084 1100 1134 1946
2004
1530 1387 1463 2541
2005
1902 1753 1858 2977
2006
2279 2139 2309
doc_102956488.pptx