Forecasting To Accompany Russell and Taylor, Operations Management, 4th Edition, 2003 Prentice-Hall, Inc. All rights reserved.
Forecasting Predicting future events Predicting future events Usually demand behavior over a time frame Usually demand behavior over a time frame
Forecast: A statement about the future value of a variable of interest such as demand. Forecasts affect decisions and activities throughout an organization –Accounting, finance –Human resources –Marketing –MIS –Operations –Product / service design
AccountingCost/profit estimates FinanceCash flow and funding Human ResourcesHiring/recruiting/training MarketingPricing, promotion, strategy MISIT/IS systems, services OperationsSchedules, MRP, workloads Product/service designNew products and services Uses of Forecasts
To help managers plan the system To help managers plan the use of the system.
Assumes causal system past ==> future Forecasts rarely perfect because of randomness Forecasts more accurate for groups vs. individuals Forecast accuracy decreases as time horizon increases I see that you will get an A this semester. Features Common to All Forecasts
Elements of a Good Forecast Timely Accurate Reliable Meaningful Written Easy to use
Time Frame in Forecasting Short-range to medium-range Short-range to medium-range Daily, weekly monthly forecasts of sales data Daily, weekly monthly forecasts of sales data Up to 2 years into the future Up to 2 years into the future Long-range Long-range Strategic planning of goals, products, markets Strategic planning of goals, products, markets Planning beyond 2 years into the future Planning beyond 2 years into the future
Steps in the Forecasting Process Step 1 Determine purpose of forecast Step 2 Establish a time horizon Step 3 Select a forecasting technique Step 4 Gather and analyze data Step 5 Prepare the forecast Step 6 Monitor the forecast “The forecast”
Forecasting Process 6. Check forecast accuracy with one or more measures 4. Select a forecast model that seems appropriate for data 5. Develop/compute forecast for period of historical data 8a. Forecast over planning horizon 9. Adjust forecast based on additional qualitative information and insight 10. Monitor results and measure forecast accuracy 8b. Select new forecast model or adjust parameters of existing model 7. Is accuracy of forecast acceptable? 1. Identify the purpose of forecast 3. Plot data and identify patterns 2. Collect historical data
Approaches to Forecasting Qualitative methodsQualitative methods –Based on subjective methods Quantitative methodsQuantitative methods –Based on mathematical formulas
Approaches to Forecasting Judgmental (Qualitative)- uses subjective inputs Time series - uses historical data assuming the future will be like the past Associative models - uses explanatory variables to predict the future
Judgmental Forecasts Executive opinions Sales force opinions Consumer surveys Outside opinion Delphi method –Opinions of managers and staff –Achieves a consensus forecast
Time Series A time series is a time-ordered sequence of observations taken at regular intervals (eg. Hourly, daily, weekly, monthly, quarterly, annually)
Demand Behavior Trend Trend gradual, long-term up or down movement gradual, long-term up or down movement Cycle Cycle up & down movement repeating over long time frame; wavelike variations of more than one year’s duration up & down movement repeating over long time frame; wavelike variations of more than one year’s duration Seasonal pattern Seasonal pattern periodic oscillation in demand which repeats; short-term regular variations in data periodic oscillation in demand which repeats; short-term regular variations in data Irregular variations caused by unusual circumstances Irregular variations caused by unusual circumstances Random movements follow no pattern; caused by chance Random movements follow no pattern; caused by chance
Forms of Forecast Movement Time (a) Trend Time (d) Trend with seasonal pattern Time (c) Seasonal pattern Time (b) Cycle Demand Demand Demand Demand Random movement
Trend Irregular variatio n Seasonal variations Cycles Forms of Forecast Movement
Time Series Methods Naive forecasts Naive forecasts Forecast = data from past period Forecast = data from past period Statistical methods using historical data Statistical methods using historical data Moving average Moving average Exponential smoothing Exponential smoothing Linear trend line Linear trend line Assume patterns will repeat Assume patterns will repeat Demand?
Naive Forecasts Uh, give me a minute.... We sold 250 wheels last week.... Now, next week we should sell.... The forecast for any period equals the previous period’s actual value.
Simple to use Virtually no cost Quick and easy to prepare Data analysis is nonexistent Easily understandable Cannot provide high accuracy Can be a standard for accuracy Naïve Forecasts
Stable time series data –F(t) = A(t-1) Seasonal variations –F(t) = A(t-n) Data with trends –F(t) = A(t-1) + (A(t-1) – A(t-2)) Uses for Naïve Forecasts
Techniques for Averaging Moving Average Weighted Moving Average Exponential Smoothing Averaging techniques smooth fluctuations in a time series.
Moving Average MA n = n i = 1 A t-i n where n =number of periods in the moving average A t- i =actual demand in period t- i Average several periods of data Average several periods of data Dampen, smooth out changes Dampen, smooth out changes Use when demand is stable with no trend or seasonal pattern Use when demand is stable with no trend or seasonal pattern
Moving Averages Moving average – A technique that averages a number of recent actual values, updated as new values become available. F t = MA n = n A t-n + … A t-2 + A t-1 Ft = Forecast for time period t MAn= n period moving average
Jan120 Feb90 Mar100 Apr75 May110 June50 July75 Aug130 Sept110 Oct90 ORDERS MONTHPER MONTH = = 110 orders for Nov Simple Moving Average F11 =MA 3
Jan120– Feb90 – Mar100 – Apr May June July Aug Sept Oct Nov –110.0 ORDERSTHREE-MONTH MONTHPER MONTHMOVING AVERAGE Simple Moving Average
Jan120– Feb90 – Mar100 – Apr May June July Aug Sept Oct Nov –110.0 ORDERSTHREE-MONTH MONTHPER MONTHMOVING AVERAGE = 91 orders for Nov Simple Moving Average F11 MA 5 =
Simple Moving Average Jan120– – Feb90 – – Mar100 – – Apr – May – June July Aug Sept Oct Nov – ORDERSTHREE-MONTHFIVE-MONTH MONTHPER MONTHMOVING AVERAGEMOVING AVERAGE
Smoothing Effects – – – – – – 0 0 – ||||||||||| JanFebMarAprMayJuneJulyAugSeptOctNov Orders Month Actual
Smoothing Effects – – – – – – 0 0 – ||||||||||| JanFebMarAprMayJuneJulyAugSeptOctNov 3-month Actual Orders Month
Smoothing Effects – – – – – – 0 0 – ||||||||||| JanFebMarAprMayJuneJulyAugSeptOctNov 5-month 3-month Actual Orders Month
Weighted Moving Average WMA n = i = 1 W i A t-i where W i = the weight for period i, between 0 and 100 percent W i = 1.00 Adjusts moving average method to more closely reflect data fluctuations Adjusts moving average method to more closely reflect data fluctuations n
Weighted Moving Averages Weighted moving average – More recent values in a series are given more weight in computing the forecast. F t = WMA n = n w n A t-n + … w n-1 A t-2 + w 1 A t-1
Weighted Moving Average Example MONTH WEIGHT DATA August 17%130 September 33%110 October 50%90 November forecast WMA 3 = 3 i = 1 Wi AiWi AiWi AiWi Ai = (0.50)(90) + (0.33)(110) + (0.17)(130) = orders
Exponential Smoothing Premise--The most recent observations might have the highest predictive value. – Therefore, we should give more weight to the more recent time periods when forecasting. F t = F t-1 + ( A t-1 - F t-1 )
Exponential Smoothing Weighted averaging method based on previous forecast plus a percentage of the forecast error A-F is the error term, is the % feedback or a percentage of forecast error F t = F t-1 + ( A t-1 - F t-1 ) Ft =forecast for the next period At -1 =actual demand for the present period Ft -1 =previously determined forecast for the present period α=weighting factor, smoothing constant α =weighting factor, smoothing constant
Exponential Smoothing F t = (1- α) F t-1 + α A t-1
Averaging method Averaging method Weights most recent data more strongly Weights most recent data more strongly Reacts more to recent changes Reacts more to recent changes Widely used, accurate method Widely used, accurate method Exponential Smoothing
Effect of Smoothing Constant 0.0 1.0 If = 0.20, then F t +1 = 0.20 A t F t If = 0, then F t +1 = 0 A t + 1 F t 0 = F t Forecast does not reflect recent data If = 1, then F t +1 = 1 A t + 0 F t = A t Forecast based only on most recent data
PERIODMONTHDEMAND 1Jan37 2Feb40 3Mar41 4Apr37 5May 45 6Jun50 7Jul 43 8Aug 47 9Sep 56 10Oct52 11Nov55 12Dec 54 Exponential Smoothing- Example 1
PERIODMONTHDEMAND 1Jan37 2Feb40 3Mar41 4Apr37 5May 45 6Jun50 7Jul 43 8Aug 47 9Sep 56 10Oct52 11Nov55 12Dec 54 F 2 = 1 + (1 - )F 1 = (0.30)(37) + (0.70)(37) = 37 F 3 = 2 + (1 - )F 2 = (0.30)(40) + (0.70)(37) = 37.9 F 13 = 12 + (1 - )F 12 = (0.30)(54) + (0.70)(50.84) = Exponential Smoothing
FORECAST, F t + 1 PERIODMONTHDEMAND( = 0.3) 1Jan37– 2Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec Jan–51.79 Exponential Smoothing
FORECAST, F t + 1 PERIODMONTHDEMAND( = 0.3)( = 0.5) 1Jan37–– 2Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec Jan– Exponential Smoothing
70 70 – – – – – – – 0 0 – ||||||||||||| Actual Orders Month Exponential Smoothing Forecasts
70 70 – – – – – – – 0 0 – ||||||||||||| Actual Orders Month = 0.30 Exponential Smoothing Forecasts
70 70 – – – – – – – 0 0 – ||||||||||||| = 0.50 Actual Orders Month = 0.30 Exponential Smoothing Forecasts
Exponential Smoothing-Example 2 Exponential Smoothing-Example 2
Picking a Smoothing Constant .1 .4 Actual
y = a + bx where a =intercept (at period 0) b =slope of the line x =the time period y =forecast for demand for period x Linear Trend Line
y = a + bx where a =intercept (at period 0) b =slope of the line x =the time period y =forecast for demand for period x b = a = y - b x where n =number of periods x == mean of the x values y == mean of the y values xy - nxy x 2 - nx 2 x n y n Linear Trend Line
Calculating a and b b = n(ty) - ty nt 2 - ( t) 2 a = y - bt n
x (PERIOD) y (DEMAND) Linear Trend Calculation Example
x (PERIOD) y (DEMAND) xyx Linear Trend Calculation Example
x (PERIOD) y (DEMAND) xyx x = = 6.5 y = = b = = = 1.72 a = y - bx = (1.72)(6.5) = (12)(6.5)(46.42) (6.5) 2 xy - nxy x 2 - nx
Linear Trend Calculation Example x (PERIOD) y (DEMAND) xyx x = = 6.5 y = = b = = = 1.72 a = y - bx = (1.72)(6.5) = (12)(6.5)(46.42) (6.5) 2 xy - nxy x 2 - nx Linear trend line y = x
Least Squares Example x (PERIOD) y (DEMAND) xyx x = = 6.5 y = = b = = = 1.72 a = y - bx = (1.72)(6.5) = (12)(6.5)(46.42) (6.5) 2 xy - nxy x 2 - nx Linear trend line y = x Forecast for period 13 y = (13) y = units
Linear Trend Line – – – – – – – 0 0 – ||||||||||||| Demand Period
Linear Trend Line – – – – – – – 0 0 – ||||||||||||| Actual Demand Period
Linear Trend Line – – – – – – – 0 0 – ||||||||||||| Actual Demand Period Linear trend line
Trend-Adjusted Exponential Smoothing A variation of simple exponential smoothing can be used when a time series exhibits trend and it is called trend-adjusted exponential smoothing or double smoothing If a series exhibits trend, and simple smoothing is used on it the forecasts will all lag the trend: if the data are increasing, each forecast will be too low; if the data are decreasing, each forecast will be too high.
Trend-Adjusted Exponential Smoothing TAF t+1 = S t + T t Where S t = Smoothed forecast T t = Current trend estimate and S t =TAF t + α (A t – TAF t ) T t = T t-1 + β (TAF t – TAF t-1 – T t-1 ) α and β are smoothing constants
Seasonal Adjustments Repetitive increase/ decrease in demand Models of seasonality: Additive (seasonality is expressed as a quantity that is added to or subtracted from the series average) Multiplicative (seasonality is expressed as a percentage of the average (or trend)amount)
Seasonal Adjustments The seasonal percentages in the multiplicative model are referred to as seasonal relatives or seasonal indexes The seasonal percentages in the multiplicative model are referred to as seasonal relatives or seasonal indexes
Seasonal Adjustments Use seasonal factor to adjust forecast Use seasonal factor to adjust forecast Seasonal factor = S i = DiDiDDDiDiDD
Seasonal Adjustment Total DEMAND (1000’S PER QUARTER) YEAR1234Total
Seasonal Adjustment Total DEMAND (1000’S PER QUARTER) YEAR1234Total S 1 = = = 0.28 D1D1DDD1D1DD S 2 = = = 0.20 D2D2DDD2D2DD S 4 = = = 0.37 D4D4DDD4D4DD S 3 = = = 0.15 D3D3DDD3D3DD
Seasonal Adjustment Total DEMAND (1000’S PER QUARTER) YEAR1234Total S i
Seasonal Adjustment Total DEMAND (1000’S PER QUARTER) YEAR1234Total S i y = x = (4) = For 2002
Seasonal Adjustment SF 1 = (S 1 ) (F 5 )SF 3 = (S 3 ) (F 5 ) = (0.28)(58.17) = 16.28= (0.15)(58.17) = 8.73 SF 2 = (S 2 ) (F 5 )SF 4 = (S 4 ) (F 5 ) = (0.20)(58.17) = 11.63= (0.37)(58.17) = Total DEMAND (1000’S PER QUARTER) YEAR1234Total S i y = x = (4) = For 2002
Centered Moving Average A commonly used method for representing the trend portion of a time series involves a centered moving average. By virtue of its centered position it looks forward and looks backward, so it is able to closely follow data movements whether they involve trends, cycles, or random variability alone.
Computing Seasonal Relatives by Using Centered Moving Averages The ratio of demand at period i to the centered average at period i is an estimate of the seasonal relative at that point.
Associative Forecasting Predictor variables - used to predict values of variable interest Regression - technique for fitting a line to a set of points Least squares line - minimizes sum of squared deviations around the line
Causal Modeling with Linear Regression Study relationship between two or more variables Study relationship between two or more variables Dependent variable y depends on independent variable x y = a + bx Dependent variable y depends on independent variable x y = a + bx
Linear Model Seems Reasonable A straight line is fitted to a set of sample points. Computed relationship
Linear Regression Formulas a = y - b x b = where a =intercept (at period 0) b =slope of the line x == mean of the x data y == mean of the y data xy - nxy x 2 - nx 2 x n y n
Linear Regression Example xy (WINS)(ATTENDANCE) xyx
Linear Regression Example xy (WINS)(ATTENDANCE) xyx x = = y = = b = = = 4.06 a = y - bx = (4.06)(6.125) = xy - nxy 2 x 2 - nx 2 (2,167.7) - (8)(6.125)(43.36) (311) - (8)(6.125) 2
Linear Regression Example xy (WINS)(ATTENDANCE) xyx x = = y = = b = = = 4.06 a = y - bx = (4.06)(6.125) = xy - nxy 2 x 2 - nx 2 (2,167.7) - (8)(6.125)(43.36) (311) - (8)(6.125) 2 y = x y = (7) = 46.88, or 46,880 Regression equation Attendance forecast for 7 wins
Linear Regression Line 60,000 60,000 – 50,000 50,000 – 40,000 40,000 – 30,000 30,000 – 20,000 20,000 – 10,000 10,000 – ||||||||||| Wins, x Attendance, y
Linear Regression Line ||||||||||| ,000 60,000 – 50,000 50,000 – 40,000 40,000 – 30,000 30,000 – 20,000 20,000 – 10,000 10,000 – Linear regression line, y = x Wins, x Attendance, y
Correlation and Coefficient of Determination Correlation, r Correlation, r Measure of strength and direction of relationship between two variables Measure of strength and direction of relationship between two variables Varies between and Varies between and Coefficient of determination, r 2 Coefficient of determination, r 2 Percentage of variation in dependent variable resulting from changes in the independent variable. Percentage of variability in the values of the dependent variable that is explained by the independent variable. Percentage of variation in dependent variable resulting from changes in the independent variable. Percentage of variability in the values of the dependent variable that is explained by the independent variable.
Computing Correlation n xy - x y [ n x 2 - ( x ) 2 ] [ n y 2 - ( y ) 2 ] r = Coefficient of determination r 2 = (0.947) 2 = r = (8)(2,167.7) - (49)(346.9) [(8)(311) - (49 )2 ] [(8)(15,224.7) - (346.9) 2 ] r = 0.947
Multiple Regression Study the relationship of demand to two or more independent variables y = 0 + 1 x 1 + 2 x 2 … + k x k where 0 =the intercept 1, …, k =parameters for the independent variables x 1, …, x k =independent variables
Important Points in Using Regression Important Points in Using Regression Always plot the data to verify that a linear relationship is appropriate Check whether the data is time- dependent. If so use time series instead of regression A small correlation may imply that other variables are important
Forecast Accuracy Error = Actual - Forecast Error = Actual - Forecast Find a method which minimizes error Find a method which minimizes error Mean Absolute Deviation (MAD) Mean Absolute Deviation (MAD) Mean Squared Error (MSE) Mean Squared Error (MSE) Mean Absolute Percent Deviation (MAPE) Mean Absolute Percent Deviation (MAPE)
Mean Absolute Deviation (MAD) where t = the period number t = the period number A t = actual demand in period t A t = actual demand in period t F t = the forecast for period t F t = the forecast for period t n = the total number of periods n = the total number of periods = the absolute value A t - F t n MAD =
MAD Example PERIODDEMAND, A t F t ( =0.3)
MAD Example –– PERIODDEMAND, A t F t ( =0.3)(A t - F t ) |A t - F t |
MAD Example –– PERIODDEMAND, D t F t ( =0.3)(D t - F t ) |D t - F t | A t - F t n MAD= = =
MAD, MSE, and MAPE MSE = Actualforecast ) n ( MAPE = Actualforecas t n / Actual*100)
Example 10
Forecast Control Reasons for out-of-control forecasts Reasons for out-of-control forecasts (sources of forecast errors) (sources of forecast errors) Change in trend Change in trend Appearance of cycle Appearance of cycle Inadequate forecasts Inadequate forecasts Irregular variations Irregular variations Incorrect use of forecasting technique Incorrect use of forecasting technique
Controlling the Forecast A forecast is deemed to perform adequately when the errors exhibit only random variations Control chart –A visual tool for monitoring forecast errors –Used to detect non-randomness in errors Forecasting errors are in control if –All errors are within the control limits –No patterns, such as trends or cycles, are present
Tracking Signal Compute each period Compute each period Compare to control limits Compare to control limits Forecast is in control if within limits Forecast is in control if within limits Use control limits of +/- 2 to +/- 5 MAD Tracking signal = = (A t - F t ) MADEMAD Bias: persistent tendency for forecasts to be greater or less than actual values
Tracking Signal Values ––– DEMANDFORECAST,ERROR E = PERIODD t F t A t - F t (A t - F t )MAD
Tracking Signal Values ––– DEMANDFORECAST,ERROR E = PERIODA t F t A t - F t (A t - F t )MAD TS 3 = = Tracking signal for period 3
Tracking Signal Values –––– DEMANDFORECAST,ERROR E =TRACKING PERIODA t F t A t - F t (A t - F t )MADSIGNAL
Tracking Signal Plot 3 3 – 2 2 – 1 1 – 0 0 – -1 -1 – -2 -2 – -3 -3 – ||||||||||||| Tracking signal (MAD) Period
Tracking Signal Plot 3 3 – 2 2 – 1 1 – 0 0 – -1 -1 – -2 -2 – -3 -3 – ||||||||||||| Tracking signal (MAD) Period Exponential smoothing ( = 0.30)
Tracking Signal Plot 3 3 – 2 2 – 1 1 – 0 0 – -1 -1 – -2 -2 – -3 -3 – ||||||||||||| Tracking signal (MAD) Period Exponential smoothing ( = 0.30) Linear trend line
Statistical Control Charts = = = = (A t - F t ) 2 n - 1 Using we can calculate statistical control limits for the forecast error Using we can calculate statistical control limits for the forecast error Control limits are typically set at 3 Control limits are typically set at 3
Statistical Control Charts Errors – – – 0 0 – – – – ||||||||||||| Period
Statistical Control Charts Errors – – – 0 0 – – – – ||||||||||||| Period UCL = +3 LCL = -3
Choosing a Forecasting Technique No single technique works in every situation Two most important factors –Cost –Accuracy Other factors include the availability of: –Historical data –Computers –Time needed to gather and analyze the data –Forecast horizon