Description
Demand forecasting in supply chain management
FORECASTING
1
CUSTOMER?S BEHAVIOR AND SUPPLY CHAIN FORECASTING
Customer?s Behavior
Earlier:
To buy particular brand Effects Supply Chain Planning & Forecasting
Now:
To buy product
Results in Interbrand competition
2
IMPORTANT ISSUES IN INTER-BRAND COMPETITION
• Manufacturers now compete less on product and quality, which are often comparable. • Availability of the product is more important. • Need is to make the product continuously. • Anticipate the need rather than react to the need.
3
IMPORTANT CONSIDERATIONS IN TODAY’S CONTEXT
• Inventory turns • Speed to market.
4
DEMAND MANAGEMENT IN SUPPLY CHAIN MANAGEMENT
Supply chain management
Marketing Customer Relationship Management
Demand Planning
Sales Forecasting Management
5
Demand planning is the driver of all Supply Chain processes...
Market demands Unknown demand (forecast based on past data)
S+OP Model
Aggregated demands
Deman d Plan Inventory Policy Information flow Order flow Physical flow
Supply Plan
Manufacturing Plan
Supplier Plan
Known demand (order)
Inventory Status
mfg. order
Customers
Customer Order
Distrib. Centres
Manufact . Replenish.
Supplier Replenish.
6
FORECASTING - 1
• Forecasting is a technique to plan the future activities and is based on the past data. • The past data is systematically put in a predetermined way to prepare estimates for the future. • Forecasting is a quantitative/qualitative technique to project the demand for a product or service.
7
Forecasting is an Integral Part of Business Planning
Inputs: Market, Economic, Other Forecast Method(s) Demand Estimates
Sales Forecast
Management Team
Business Strategy
Production Resource Forecasts
8
CHARACTERISTICS OF FORECASTS -1
• Forecasts are always inaccurate and include both the expected value of the forecast and a measure of forecast error.
– An average sale of an item is expected in the range of 100 to 1900 and 900 to 1100 is 1000. – Forecasting error (or demand uncertainty) is important for decision making.
• Long term forecasts are usually less accurate than short-term forecasts.
9
CHARACTERISTICS OF FORECASTS -2
• Aggregate forecasts are usually far more accurate than disaggregate forecasts as they tend to have a smaller standard deviation of error relative to the mean. • In general, the further up the supplier chain a company is (or further they are from the consumer), the greater the distortion of information they receive.
10
FACTORS RELATED TO THE DEMAND FORECAST
• • • • • • Past demand. Lead time of product. Planned advertising or marketing efforts. State of economy. Planned price discounts. Actions taken by the competitors.
11
Forecasting Methods
• Qualitative Approaches • Quantitative Approaches
12
Methods of demand forecasting
Demand Forecasting
Qualitative Analysis
Quantitative Analysis
Customer Survey
Sales Force Composite
Time Series Analysis
Causal Analysis
Executive Opinion
Delphi Method
Simple Average
Weighted Moving Average
Trend Analysis
Past Analogy
Simple Moving Average
Simple Exponential Smoothing
Winters’s Triple Exponential Smoothing
Holt’s Double Exponential Smoothing
Forecast by Linear Regression Analysis
13
FORECASTING APPROACHES
Broadly forecasting approaches can be divided into three categories. Different models are used under each category. The main classification is as follows: • Qualitative approaches: Frequently used for longerrange strategic planning and facilities decision. • Quantitative approaches: Frequently used for shortterm operational planning such as production and inventory control. This approach uses more analytical method, i.e. time series analysis models. • Casual Technique approach: These approaches are helpful in intermediate-term aggregate planning for variety of planning situations.
14
Qualitative Approaches
• Usually based on judgments about causal factors that underlie the demand of particular products or services • Do not require a demand history for the product or service, therefore are useful for new products/services • Approaches vary in sophistication from scientifically conducted surveys to intuitive hunches about future events • The approach/method that is appropriate 15 depends on a product?s life cycle stage
Qualitative Methods
• • • • • • • Educated guess intuitive hunches Executive committee consensus Delphi method Survey of sales force Survey of customers Historical analogy Market research scientifically conducted surveys
16
Quantitative Forecasting Approaches
• Based on the assumption that the “forces” that generated the past demand will generate the future demand, i.e., history will tend to repeat itself • Analysis of the past demand pattern provides a good basis for forecasting future demand • Majority of quantitative approaches fall in the category of time series analysis
17
Time Series Analysis
• A time series is a set of numbers where the order or sequence of the numbers is important, e.g., historical demand • Analysis of the time series identifies patterns • Once the patterns are identified, they can be used to develop a forecast
18
Components of a Time Series
• Trends are noted by an upward or downward sloping line. • Cycle is a data pattern that may cover several years before it repeats itself. • Seasonality is a data pattern that repeats itself over the period of one year or less. • Random fluctuation (noise) results from random variation or unexplained causes.
19
Quantitative Forecasting Approaches
• • • • • Linear Regression Simple Average Simple Moving Average Weighted Moving Average Exponential Smoothing (exponentially weighted moving average) • Exponential Smoothing with Trend (double exponential smoothing)
20
Long-Range Forecasts
• Time spans usually greater than one year • Necessary to support strategic decisions about planning products, processes, and facilities
21
Simple Linear Regression
• Linear regression analysis establishes a relationship between a dependent variable and one or more independent variables. • In simple linear regression analysis there is only one independent variable. • If the data is a time series, the independent variable is the time period. • The dependent variable is whatever we wish to forecast.
22
Simple Linear Regression
• Regression Equation This model is of the form: Y = a + bX
Y = dependent variable X = independent variable a = y-axis intercept b = slope of regression line
23
Simple Linear Regression
• Constants a and b The constants a and b are computed using the following equations:
a= x2 ? y-? x? xy ? n ? x2 -( ? x)2 n? xy-? x? y n ? x2 -( ? x)2
b=
24
Simple Linear Regression
• Once the a and b values are computed, a future value of X can be entered into the regression equation and a corresponding value of Y (the forecast) can be calculated.
25
Short-Range Forecasting Methods
• • • • • Simple average (Simple) Moving Average Weighted Moving Average Exponential Smoothing Exponential Smoothing with Trend
26
FORECASTING BY USING SIMPLE AVERAGE METHOD
The simple average is calculated as under: Sum of demands for all past periods Simple Average = ____________________________ Number of demand periods D1 + D2 + D3 ………+ Dn =__________________________ n Where D1 = Demand of period No.1, i.e. the most recent period D2 = Demand of period No.2, i.e. the period just prior to recent period D3 = Demand of period No.3, i.e. the period just prior to period No.2 Dn = Demand of period No.n
27
EXERCISE
• The demand for 100 Watt bulbs in the months of January, February, March and April has been 500, 600, 800 and 700. forecast the monthly demand for the bulbs.
Solution
D1 + D2 + D3 + D4 Simple Average (SA) = ________________ n 500 + 600 + 800 + 700 = ___________________ 4 2600 = ___________ 4 = 650 The average demand for 100 Watt bulbs is 650 per month.
28
Simple Moving Average
• An averaging period (AP) is given or selected • The forecast for the next period is the arithmetic average of the AP most recent actual demands • It is called a “simple” average because each period used to compute the average is equally weighted • . . . more
29
Simple Moving Average
• It is called “moving” because as new demand data becomes available, the oldest data is not used • By increasing the AP, the forecast is less responsive to fluctuations in demand (low impulse response and high noise dampening) • By decreasing the AP, the forecast is more responsive to fluctuations in demand (high impulse response and low noise 30 dampening)
FORECASTING BY USING SIMPLE MOVING AVAERAGE
Sum of demands for last n periods Moving Average (MA) = ______________________________ Number of periods selected in the moving average n ? Dt t=1 MA = _________ = 1/n x D1 +1/n x D2 +..... +1/n x Dn n Where t = 1, the earliest period in the n-period average t = n is the most recent time period • If we have a demand data of 8 months and have selected a 5 months period to work out the simple moving average, then:
31
D1 + D2 + D3 + D4 + D5
• • • • • • • • • • • •
MA1 =
____________________ 5 D2 + D3 + D4 + D5 + D6 MA2 = ____________________ 5 D3 + D4 + D5 + D6 + D7 MA3 = ___________________ 5 D4 + D5 + D6 + D7 + D8 MA4 = ___________________ 5 Based on the eight months data, these averages can be plotted on the graph to provide the trend for the demand.
32
EXERCISE FOR FORECASTING USING MOVING AVERAGE
• The demand for 100 Watt bulbs in the past 8 months is given as below: Month Demand • January 500 • February 600 • March 800 • April 700 • May 700 • June 800 • July 600 • August 500 Calculate the moving average for a period of 5 months.
33
SOLUTION FORE MOVING AVERAGE
• MA1 = 500 + 600 + 800 + 700 + 700 ________________________ = 660 5 600 + 800 + 700 + 700 + 800 ________________________ = 720 5 800 + 700 + 700 + 800 + 600 _________________________= 720 5 700 + 700 + 800 + 600 + 500 _________________________ = 660 5
34
• MA2 =
• MA3 =
• MA4 =
Computer Software for Forecasting
• Examples of computer software with forecasting capabilities
Primarily for – Forecast Pro forecasting – Autobox – SmartForecasts for Windows Have – SAS Forecasting – SPSS modules – SAP – POM Software Libary
35
Forecasting in Small Businesses and Start-Up Ventures
• Forecasting for these businesses can be difficult for the following reasons:
– Not enough personnel with the time to forecast – Personnel lack the necessary skills to develop good forecasts – Such businesses are not data-rich environments – Forecasting for new products/services is always difficult, even for the experienced 36 forecaster
Measurement of Forecasting Errors
• • • • • Running Sum of Forecast Errors (RSFE) Mean Forecast Error (MFE) Mean Absolute Deviation (MAD) Mean Squared Error (MSE) Mean Absolute Percentage Error (MAPE) Tracking Signal (TS)
37
•
MEAN ABSOLUTE DEVIATION (MAD)
• • • MAD is the ratio of sum of absolute deviations for all periods to the total number of periods studied. It is represented as below: Sum of absolute values of deviations for all periods MAD = Total number of periods studied n ? | Forecasted demand – Actual demand |i i=1 = ____________________________________ n where n is the number of periods studied. Actual demand is compared with the forecasted demand for each period (i). When the forecast is accurate, actual demand equals to the forecasted demand and there is no error.
38
•
BIAS
Sum of algebraic errors of all the periods
• Bias
=
______________________________________ Total number of periods studied n ? (Forecasted demand – Actual demand) i i=1 = __________________________________
n • Bias indicates the directional tendency of forecast errors.
39
GOOD FORECAST - 1
• The ideal forecast should have zero MAD and zero Bias. • Usually trade off is attempted between MAD and Bias i.e. one must be kept low at the cost of the other. In general, focus should be on MAD. • Lowering MAD to or near zero will automatically hold Bias low.
40
GOOD FORECAST - 2
The essentials for effective forecasts are: • It should be accurate enough to help the decision making process. • It should provide timely indications of major shifts in process performance. • It should be simple in use. • It should be easily understandable.
41
EXCERCISE
• The demand for sewing machine was estimated as 1000 per month for each of 5 months. Later on the actual demand was found as 900, 1050, 1000, 1100 and 950 respectively. Workout MAD and Bias. Analyze if the forecast made is accurate.
42
SOLUTION
|1000–900| + |1000-1050| + |1000-1000| + |1000-1100| + |1000-950| • MAD = _________________________________________________ 5 100 + 50 + 0 +100 + 50 _____________________ 5 = 60 units of sewing machines. (1000–900) + (1000-1050) + (1000-1000) + (1000-1100) + (1000-950)
=
Bias = __________________________________________________________ 5 • 100 - 50 + 0 -100 + 50 • = ___________________ • 5 • = 0 units of sewing machines.
43
• In this case, MAD is 60 units, • Whereas Bias has no deviation. • Since MAD measures the overall accuracy of the forecasting method, it is found that:
– The forecast is not based on accurate model and the error is 6% (60/1000 x 100).
44
TRACKING SIGNALS (TS)
• The difference in forecasted demand and the actual demand should be as low as possible. • Tracking Signals are often used to monitor the forecasts especially when the overall forecast is suspect. • If the TS is around zero, the forecasting model is performing well. • A forecast is considered out of control, if the value of Tracking Signal exceeds plus 45 or minus 4.
TRACKING SIGNALS (TS)
• The TS indicates the direction of the forecasting error. • If TS is positive – increase the forecasts, but if it is negative – decrease the forecasts. • It is the ratio of the cumulative algebraic sum of the deviations between the forecasts and the actual values to the mean absolute deviation.
46
TRACKING SIGNAL
Tracking Signals (TS) = Algebraic sum of deviations ___________________________ Mean Absolute Deviations (MAD) n ? (Actual demand i – Forecast demand i) i = 1_____________________________ MAD
=
•
47
Forecast Control Limits
Forecast Error 0 + 3. s Upper Control Limit (UCL) Central Line (CL) Lower Control Limit (LCL)
Targeted or Aimedat Mean Forecast Error = 0 0 - 3. s
Time
Forecast Control Limits
48
ADVATNTAGES OF FORECASTING - 1
• Past data provides guidance for future and is a tool for training. Forecasts based on past data helps in correct planning. • Forecasting of customer’s demand helps in strategy planning, capacity planning, location planning and layout planning. • Past data provides trends, which is used for forecasting the future trends and helps in deciding on products or services to be pursued or to be stopped or abandoned. • Forecasting of manufacturing is essential to ensure availability of materials for sub-assemblies and final assemblies.
49
ADVATNTAGES OF FORECASTING - 2
• Forecasting helps in optimizing various costs as it lays down bench marks to control the project. Actual demand and actual output are monitored, compared with the previous plans and feedback into demand forecasting sub-system. • Forecasting by specifying future demands can reduce the costs of readjustment of operations in response to unexpected deviation. • Accurate estimation of future demands of goods and services through forecasting increases operating efficiency. • Forecasting is an important component of strategic and operational planning. • Utilization of the plant can be improved with correct forecasts.
50
Demand management - Quiz
1. List four data sources that a forecasting system may use. 2. What forecast strategy is most appropriate for low value inventory items with low sales volume? 3. List three ways in which demand can be segmented in a good forecasting system. 4. Is make-to-order appropriate for a manufacturer with a long lead-time process?
51
END
52
Demand management - Answers
1. List four data sources that a forecasting system may use.
• Market data, company sales data, customer EPOS data, and customer stock data.
2. What forecast strategy is most appropriate for low value inventory items with low sales volume?
• Re-order point with minimum and maximum stock levels.
3. List three ways in which demand can be segmented in a good forecasting system.
• By market, by product, by customer.
4. Is make-to-order appropriate for a manufacturer with a long lead-time process?
• No - make to forecast is appropriate if the lead-time is longer than the time between order receipt and required shipment.
53
Sources of Forecasting Data and Help
• Government agencies at the local, regional, state, and federal levels • Industry associations • Consulting companies
54
CHALLENGES IN DESIGN OF SUPPLY CHAIN
Supply chain design should be:
– Robust yet sensitive – Strong yet agile / Quick-moving. – Exhaustive yet responsive
55
Forecasting is an essential first step
• By definition or experience, the forecast is always wrong. The question is: By how much? and does it matter? Best practice forecasting uses various statistical algorithms and a wide variety of data sources – Market data, company sales data, customer Expected Point Of Supply (EPOS) data, and customer stock data if available There are strategies that can be used to achieve effective demand planning, even if demand appears to be unpredictable
100% Forecast
•
•
1 2 wk wks
Plan actual orders
X months
Plan to forecast
Capacity
Schedule
12 months
Time
56
Segmentation and differentiation is the key to effective demand planning...
Customer segmentation Customer Importance Market segmentation
H
H Professional Products
Consumer Products L Whole sal e Retail Direct Whole Retail Direct sal e Inventory segmentation of products H Volume
Best Sellers New Products Mktg. Campaigns Steady Sellers Forecast at item level by warehouse or plant Re-order point min/max system Forecast at customer level Automatic stock replenishment Aggregate customer forecast by line item by warehouse or plant
L
Market response times H Predictability
Ad hoc
L
Fast
Slow
L
L
Value
H 57
Inventory management converts demand planning into actions
• The total inventory (from supply commitments through to finished goods) needs to be managed as a whole, rather than as independent elements • Inventory should be proactively optimised throughout the chain, following agreed rules and policies determined by segmentation and Pareto analysis
Cumulative Sales Value
100 %
50%
0% 0% 50% 100%
Cumulative Orders
58
Example from consumer appliances industry: Inventory segmentation can be used to drive manufacturing strategy
Number of orders 80% 15% 5%
80% Sales Volume 15%
Predictable
?low/JIT component inventory ?make to forecast ?finished goods inventory depends on manufacturing cycle
5%
Unpredictable
?larger component inventory ?make to order ?no finished goods inventory (this depends on manufacturing cycle)
59
Demand Planning - What does „Good? look like ?
•Products/Customers/Markets are clearly segmented and differentiated •Inventory of materials and finished goods is deployed according to analysis of volume and frequency of usage (Pareto/ABC analysis) •Customer/channel/product profitability is understood and allocation rules are in place •Forecasting combines statistical modelling and human assumptions, and is used to drive longer lead-time processes •Sales, marketing, manufacturing, purchasing, and planning personnel are all involved in the forecasting process •Forecasts are formally “owned“ by all departments and forecast accuracy is a primary KPI •Outside the order horizon, inventory should be planned and deployed to forecast. Within the order horizon, make to order should be the goal
60
Reasons for demand forecasting
To maximize gains from events, which are the results of actions taken by the organization
To maximize gains from events external to the organization (from the external environment) To minimize losses associated with uncontrollable events external to the organization Reasons for Demand Forecasting
To develop policies that apply to people who are not part of the organization
To develop administrative plans and policy internal to an organization (e.g. personnel or budget)
To offset the actions of competitor organizations
As an input to Aggregate Production Planning and / or Material Requirements Planning (MRP)
In order to perform adequate staffing to support production requirements
In decision making for Facility Capacity Planning and for Capital Budgeting
61
doc_921944137.ppt
Demand forecasting in supply chain management
FORECASTING
1
CUSTOMER?S BEHAVIOR AND SUPPLY CHAIN FORECASTING
Customer?s Behavior
Earlier:
To buy particular brand Effects Supply Chain Planning & Forecasting
Now:
To buy product
Results in Interbrand competition
2
IMPORTANT ISSUES IN INTER-BRAND COMPETITION
• Manufacturers now compete less on product and quality, which are often comparable. • Availability of the product is more important. • Need is to make the product continuously. • Anticipate the need rather than react to the need.
3
IMPORTANT CONSIDERATIONS IN TODAY’S CONTEXT
• Inventory turns • Speed to market.
4
DEMAND MANAGEMENT IN SUPPLY CHAIN MANAGEMENT
Supply chain management
Marketing Customer Relationship Management
Demand Planning
Sales Forecasting Management
5
Demand planning is the driver of all Supply Chain processes...
Market demands Unknown demand (forecast based on past data)
S+OP Model
Aggregated demands
Deman d Plan Inventory Policy Information flow Order flow Physical flow
Supply Plan
Manufacturing Plan
Supplier Plan
Known demand (order)
Inventory Status
mfg. order
Customers
Customer Order
Distrib. Centres
Manufact . Replenish.
Supplier Replenish.
6
FORECASTING - 1
• Forecasting is a technique to plan the future activities and is based on the past data. • The past data is systematically put in a predetermined way to prepare estimates for the future. • Forecasting is a quantitative/qualitative technique to project the demand for a product or service.
7
Forecasting is an Integral Part of Business Planning
Inputs: Market, Economic, Other Forecast Method(s) Demand Estimates
Sales Forecast
Management Team
Business Strategy
Production Resource Forecasts
8
CHARACTERISTICS OF FORECASTS -1
• Forecasts are always inaccurate and include both the expected value of the forecast and a measure of forecast error.
– An average sale of an item is expected in the range of 100 to 1900 and 900 to 1100 is 1000. – Forecasting error (or demand uncertainty) is important for decision making.
• Long term forecasts are usually less accurate than short-term forecasts.
9
CHARACTERISTICS OF FORECASTS -2
• Aggregate forecasts are usually far more accurate than disaggregate forecasts as they tend to have a smaller standard deviation of error relative to the mean. • In general, the further up the supplier chain a company is (or further they are from the consumer), the greater the distortion of information they receive.
10
FACTORS RELATED TO THE DEMAND FORECAST
• • • • • • Past demand. Lead time of product. Planned advertising or marketing efforts. State of economy. Planned price discounts. Actions taken by the competitors.
11
Forecasting Methods
• Qualitative Approaches • Quantitative Approaches
12
Methods of demand forecasting
Demand Forecasting
Qualitative Analysis
Quantitative Analysis
Customer Survey
Sales Force Composite
Time Series Analysis
Causal Analysis
Executive Opinion
Delphi Method
Simple Average
Weighted Moving Average
Trend Analysis
Past Analogy
Simple Moving Average
Simple Exponential Smoothing
Winters’s Triple Exponential Smoothing
Holt’s Double Exponential Smoothing
Forecast by Linear Regression Analysis
13
FORECASTING APPROACHES
Broadly forecasting approaches can be divided into three categories. Different models are used under each category. The main classification is as follows: • Qualitative approaches: Frequently used for longerrange strategic planning and facilities decision. • Quantitative approaches: Frequently used for shortterm operational planning such as production and inventory control. This approach uses more analytical method, i.e. time series analysis models. • Casual Technique approach: These approaches are helpful in intermediate-term aggregate planning for variety of planning situations.
14
Qualitative Approaches
• Usually based on judgments about causal factors that underlie the demand of particular products or services • Do not require a demand history for the product or service, therefore are useful for new products/services • Approaches vary in sophistication from scientifically conducted surveys to intuitive hunches about future events • The approach/method that is appropriate 15 depends on a product?s life cycle stage
Qualitative Methods
• • • • • • • Educated guess intuitive hunches Executive committee consensus Delphi method Survey of sales force Survey of customers Historical analogy Market research scientifically conducted surveys
16
Quantitative Forecasting Approaches
• Based on the assumption that the “forces” that generated the past demand will generate the future demand, i.e., history will tend to repeat itself • Analysis of the past demand pattern provides a good basis for forecasting future demand • Majority of quantitative approaches fall in the category of time series analysis
17
Time Series Analysis
• A time series is a set of numbers where the order or sequence of the numbers is important, e.g., historical demand • Analysis of the time series identifies patterns • Once the patterns are identified, they can be used to develop a forecast
18
Components of a Time Series
• Trends are noted by an upward or downward sloping line. • Cycle is a data pattern that may cover several years before it repeats itself. • Seasonality is a data pattern that repeats itself over the period of one year or less. • Random fluctuation (noise) results from random variation or unexplained causes.
19
Quantitative Forecasting Approaches
• • • • • Linear Regression Simple Average Simple Moving Average Weighted Moving Average Exponential Smoothing (exponentially weighted moving average) • Exponential Smoothing with Trend (double exponential smoothing)
20
Long-Range Forecasts
• Time spans usually greater than one year • Necessary to support strategic decisions about planning products, processes, and facilities
21
Simple Linear Regression
• Linear regression analysis establishes a relationship between a dependent variable and one or more independent variables. • In simple linear regression analysis there is only one independent variable. • If the data is a time series, the independent variable is the time period. • The dependent variable is whatever we wish to forecast.
22
Simple Linear Regression
• Regression Equation This model is of the form: Y = a + bX
Y = dependent variable X = independent variable a = y-axis intercept b = slope of regression line
23
Simple Linear Regression
• Constants a and b The constants a and b are computed using the following equations:
a= x2 ? y-? x? xy ? n ? x2 -( ? x)2 n? xy-? x? y n ? x2 -( ? x)2
b=
24
Simple Linear Regression
• Once the a and b values are computed, a future value of X can be entered into the regression equation and a corresponding value of Y (the forecast) can be calculated.
25
Short-Range Forecasting Methods
• • • • • Simple average (Simple) Moving Average Weighted Moving Average Exponential Smoothing Exponential Smoothing with Trend
26
FORECASTING BY USING SIMPLE AVERAGE METHOD
The simple average is calculated as under: Sum of demands for all past periods Simple Average = ____________________________ Number of demand periods D1 + D2 + D3 ………+ Dn =__________________________ n Where D1 = Demand of period No.1, i.e. the most recent period D2 = Demand of period No.2, i.e. the period just prior to recent period D3 = Demand of period No.3, i.e. the period just prior to period No.2 Dn = Demand of period No.n
27
EXERCISE
• The demand for 100 Watt bulbs in the months of January, February, March and April has been 500, 600, 800 and 700. forecast the monthly demand for the bulbs.
Solution
D1 + D2 + D3 + D4 Simple Average (SA) = ________________ n 500 + 600 + 800 + 700 = ___________________ 4 2600 = ___________ 4 = 650 The average demand for 100 Watt bulbs is 650 per month.
28
Simple Moving Average
• An averaging period (AP) is given or selected • The forecast for the next period is the arithmetic average of the AP most recent actual demands • It is called a “simple” average because each period used to compute the average is equally weighted • . . . more
29
Simple Moving Average
• It is called “moving” because as new demand data becomes available, the oldest data is not used • By increasing the AP, the forecast is less responsive to fluctuations in demand (low impulse response and high noise dampening) • By decreasing the AP, the forecast is more responsive to fluctuations in demand (high impulse response and low noise 30 dampening)
FORECASTING BY USING SIMPLE MOVING AVAERAGE
Sum of demands for last n periods Moving Average (MA) = ______________________________ Number of periods selected in the moving average n ? Dt t=1 MA = _________ = 1/n x D1 +1/n x D2 +..... +1/n x Dn n Where t = 1, the earliest period in the n-period average t = n is the most recent time period • If we have a demand data of 8 months and have selected a 5 months period to work out the simple moving average, then:
31
D1 + D2 + D3 + D4 + D5
• • • • • • • • • • • •
MA1 =
____________________ 5 D2 + D3 + D4 + D5 + D6 MA2 = ____________________ 5 D3 + D4 + D5 + D6 + D7 MA3 = ___________________ 5 D4 + D5 + D6 + D7 + D8 MA4 = ___________________ 5 Based on the eight months data, these averages can be plotted on the graph to provide the trend for the demand.
32
EXERCISE FOR FORECASTING USING MOVING AVERAGE
• The demand for 100 Watt bulbs in the past 8 months is given as below: Month Demand • January 500 • February 600 • March 800 • April 700 • May 700 • June 800 • July 600 • August 500 Calculate the moving average for a period of 5 months.
33
SOLUTION FORE MOVING AVERAGE
• MA1 = 500 + 600 + 800 + 700 + 700 ________________________ = 660 5 600 + 800 + 700 + 700 + 800 ________________________ = 720 5 800 + 700 + 700 + 800 + 600 _________________________= 720 5 700 + 700 + 800 + 600 + 500 _________________________ = 660 5
34
• MA2 =
• MA3 =
• MA4 =
Computer Software for Forecasting
• Examples of computer software with forecasting capabilities
Primarily for – Forecast Pro forecasting – Autobox – SmartForecasts for Windows Have – SAS Forecasting – SPSS modules – SAP – POM Software Libary
35
Forecasting in Small Businesses and Start-Up Ventures
• Forecasting for these businesses can be difficult for the following reasons:
– Not enough personnel with the time to forecast – Personnel lack the necessary skills to develop good forecasts – Such businesses are not data-rich environments – Forecasting for new products/services is always difficult, even for the experienced 36 forecaster
Measurement of Forecasting Errors
• • • • • Running Sum of Forecast Errors (RSFE) Mean Forecast Error (MFE) Mean Absolute Deviation (MAD) Mean Squared Error (MSE) Mean Absolute Percentage Error (MAPE) Tracking Signal (TS)
37
•
MEAN ABSOLUTE DEVIATION (MAD)
• • • MAD is the ratio of sum of absolute deviations for all periods to the total number of periods studied. It is represented as below: Sum of absolute values of deviations for all periods MAD = Total number of periods studied n ? | Forecasted demand – Actual demand |i i=1 = ____________________________________ n where n is the number of periods studied. Actual demand is compared with the forecasted demand for each period (i). When the forecast is accurate, actual demand equals to the forecasted demand and there is no error.
38
•
BIAS
Sum of algebraic errors of all the periods
• Bias
=
______________________________________ Total number of periods studied n ? (Forecasted demand – Actual demand) i i=1 = __________________________________
n • Bias indicates the directional tendency of forecast errors.
39
GOOD FORECAST - 1
• The ideal forecast should have zero MAD and zero Bias. • Usually trade off is attempted between MAD and Bias i.e. one must be kept low at the cost of the other. In general, focus should be on MAD. • Lowering MAD to or near zero will automatically hold Bias low.
40
GOOD FORECAST - 2
The essentials for effective forecasts are: • It should be accurate enough to help the decision making process. • It should provide timely indications of major shifts in process performance. • It should be simple in use. • It should be easily understandable.
41
EXCERCISE
• The demand for sewing machine was estimated as 1000 per month for each of 5 months. Later on the actual demand was found as 900, 1050, 1000, 1100 and 950 respectively. Workout MAD and Bias. Analyze if the forecast made is accurate.
42
SOLUTION
|1000–900| + |1000-1050| + |1000-1000| + |1000-1100| + |1000-950| • MAD = _________________________________________________ 5 100 + 50 + 0 +100 + 50 _____________________ 5 = 60 units of sewing machines. (1000–900) + (1000-1050) + (1000-1000) + (1000-1100) + (1000-950)
=
Bias = __________________________________________________________ 5 • 100 - 50 + 0 -100 + 50 • = ___________________ • 5 • = 0 units of sewing machines.
43
• In this case, MAD is 60 units, • Whereas Bias has no deviation. • Since MAD measures the overall accuracy of the forecasting method, it is found that:
– The forecast is not based on accurate model and the error is 6% (60/1000 x 100).
44
TRACKING SIGNALS (TS)
• The difference in forecasted demand and the actual demand should be as low as possible. • Tracking Signals are often used to monitor the forecasts especially when the overall forecast is suspect. • If the TS is around zero, the forecasting model is performing well. • A forecast is considered out of control, if the value of Tracking Signal exceeds plus 45 or minus 4.
TRACKING SIGNALS (TS)
• The TS indicates the direction of the forecasting error. • If TS is positive – increase the forecasts, but if it is negative – decrease the forecasts. • It is the ratio of the cumulative algebraic sum of the deviations between the forecasts and the actual values to the mean absolute deviation.
46
TRACKING SIGNAL
Tracking Signals (TS) = Algebraic sum of deviations ___________________________ Mean Absolute Deviations (MAD) n ? (Actual demand i – Forecast demand i) i = 1_____________________________ MAD
=
•
47
Forecast Control Limits
Forecast Error 0 + 3. s Upper Control Limit (UCL) Central Line (CL) Lower Control Limit (LCL)
Targeted or Aimedat Mean Forecast Error = 0 0 - 3. s
Time
Forecast Control Limits
48
ADVATNTAGES OF FORECASTING - 1
• Past data provides guidance for future and is a tool for training. Forecasts based on past data helps in correct planning. • Forecasting of customer’s demand helps in strategy planning, capacity planning, location planning and layout planning. • Past data provides trends, which is used for forecasting the future trends and helps in deciding on products or services to be pursued or to be stopped or abandoned. • Forecasting of manufacturing is essential to ensure availability of materials for sub-assemblies and final assemblies.
49
ADVATNTAGES OF FORECASTING - 2
• Forecasting helps in optimizing various costs as it lays down bench marks to control the project. Actual demand and actual output are monitored, compared with the previous plans and feedback into demand forecasting sub-system. • Forecasting by specifying future demands can reduce the costs of readjustment of operations in response to unexpected deviation. • Accurate estimation of future demands of goods and services through forecasting increases operating efficiency. • Forecasting is an important component of strategic and operational planning. • Utilization of the plant can be improved with correct forecasts.
50
Demand management - Quiz
1. List four data sources that a forecasting system may use. 2. What forecast strategy is most appropriate for low value inventory items with low sales volume? 3. List three ways in which demand can be segmented in a good forecasting system. 4. Is make-to-order appropriate for a manufacturer with a long lead-time process?
51
END
52
Demand management - Answers
1. List four data sources that a forecasting system may use.
• Market data, company sales data, customer EPOS data, and customer stock data.
2. What forecast strategy is most appropriate for low value inventory items with low sales volume?
• Re-order point with minimum and maximum stock levels.
3. List three ways in which demand can be segmented in a good forecasting system.
• By market, by product, by customer.
4. Is make-to-order appropriate for a manufacturer with a long lead-time process?
• No - make to forecast is appropriate if the lead-time is longer than the time between order receipt and required shipment.
53
Sources of Forecasting Data and Help
• Government agencies at the local, regional, state, and federal levels • Industry associations • Consulting companies
54
CHALLENGES IN DESIGN OF SUPPLY CHAIN
Supply chain design should be:
– Robust yet sensitive – Strong yet agile / Quick-moving. – Exhaustive yet responsive
55
Forecasting is an essential first step
• By definition or experience, the forecast is always wrong. The question is: By how much? and does it matter? Best practice forecasting uses various statistical algorithms and a wide variety of data sources – Market data, company sales data, customer Expected Point Of Supply (EPOS) data, and customer stock data if available There are strategies that can be used to achieve effective demand planning, even if demand appears to be unpredictable
100% Forecast
•
•
1 2 wk wks
Plan actual orders
X months
Plan to forecast
Capacity
Schedule
12 months
Time
56
Segmentation and differentiation is the key to effective demand planning...
Customer segmentation Customer Importance Market segmentation
H
H Professional Products
Consumer Products L Whole sal e Retail Direct Whole Retail Direct sal e Inventory segmentation of products H Volume
Best Sellers New Products Mktg. Campaigns Steady Sellers Forecast at item level by warehouse or plant Re-order point min/max system Forecast at customer level Automatic stock replenishment Aggregate customer forecast by line item by warehouse or plant
L
Market response times H Predictability
Ad hoc
L
Fast
Slow
L
L
Value
H 57
Inventory management converts demand planning into actions
• The total inventory (from supply commitments through to finished goods) needs to be managed as a whole, rather than as independent elements • Inventory should be proactively optimised throughout the chain, following agreed rules and policies determined by segmentation and Pareto analysis
Cumulative Sales Value
100 %
50%
0% 0% 50% 100%
Cumulative Orders
58
Example from consumer appliances industry: Inventory segmentation can be used to drive manufacturing strategy
Number of orders 80% 15% 5%
80% Sales Volume 15%
Predictable
?low/JIT component inventory ?make to forecast ?finished goods inventory depends on manufacturing cycle
5%
Unpredictable
?larger component inventory ?make to order ?no finished goods inventory (this depends on manufacturing cycle)
59
Demand Planning - What does „Good? look like ?
•Products/Customers/Markets are clearly segmented and differentiated •Inventory of materials and finished goods is deployed according to analysis of volume and frequency of usage (Pareto/ABC analysis) •Customer/channel/product profitability is understood and allocation rules are in place •Forecasting combines statistical modelling and human assumptions, and is used to drive longer lead-time processes •Sales, marketing, manufacturing, purchasing, and planning personnel are all involved in the forecasting process •Forecasts are formally “owned“ by all departments and forecast accuracy is a primary KPI •Outside the order horizon, inventory should be planned and deployed to forecast. Within the order horizon, make to order should be the goal
60
Reasons for demand forecasting
To maximize gains from events, which are the results of actions taken by the organization
To maximize gains from events external to the organization (from the external environment) To minimize losses associated with uncontrollable events external to the organization Reasons for Demand Forecasting
To develop policies that apply to people who are not part of the organization
To develop administrative plans and policy internal to an organization (e.g. personnel or budget)
To offset the actions of competitor organizations
As an input to Aggregate Production Planning and / or Material Requirements Planning (MRP)
In order to perform adequate staffing to support production requirements
In decision making for Facility Capacity Planning and for Capital Budgeting
61
doc_921944137.ppt