Forecasting MD707 Operations Management Professor Joy Field
Components of the Forecast 2
Forecasting using Judgment Methods Sales force estimates Executive opinion Market research Delphi method 3
Forecasting using Time Series Methods Naïve forecasts Moving averages Weighted moving averages Exponential smoothing Trend-adjusted exponential smoothing Multiplicative seasonal method 4
Moving Average Method Use a 3-month moving average, what is the forecast for month 5? If the actual demand for month 5 is 805 customers, what is the forecast for month 6? 5 MonthCustomers
Comparison of Three-Week and Six-Week Moving Average Forecasts 6
Weighted Moving Average Method Let Calculate the forecast for Month 5. If the actual number of customers in month 5 is 805, what is the forecast for month 6? 7 MonthCustomers
Exponential Smoothing Suppose What is the forecast for Month 5? If the actual number of customers in month 5 is 805, what is the forecast for month 6? 8 MonthCustomers
Trend-Adjusted Exponential Smoothing Using months 1-4, an initial estimate of the trend for Month 5 is 2 [(4-2+4)/3 = 2]. The starting forecast for month 5 is 54+2 = 56. Using and forecast the number of customers in month 6. 9 Month Customers
Trend-Adjusted Exponential Smoothing (cont.) If the actual number of customers in month 6 is 58, what is the forecast for month 7? 10
Multiplicative Seasonal Method Procedure Calculate the trend line based on the available data using regression. Calculate the centered moving average, with the number of periods equal to the number of seasons. Calculate the seasonal relative for a period by dividing the actual demand for the period by the corresponding centered moving average. Calculate the overall estimated seasonal relative by averaging the seasonal relatives from the same periods over the cycle. Calculate the trend values for each of the periods to be forecast based on the trend line determined in Step 1. To get a forecast for a given period in a future cycle, multiply the seasonal factor by the trend values. 11
Multiplicative Seasonal Method Example QuarterDemandCMA (4 seasons)MA (2 periods) Seasonal Relatives Normalized S.R. Year 1, Q1 100 Year 1, Q Year 1, Q Year 1, Q Year 2, Q Year 2, Q Year 2, Q3 384 Total3.924 Year 2, Q4 216 Year 3, Q1331(trend value*)227(forecast) Year 3, Q2344(trend value*)480(forecast) Year 3, Q3356(trend value*)417(forecast) Year 3, Q4369(trend value*)275(forecast) 12 * Using regression, the trend line is t.
Linear Regression where y = dependent (predicted) variable x = independent (predictor) variable a = y-intercept of the line (i.e., value of y when x = 0) b = slope of the line 13 y = a + bx
Linear Regression Line Relative to Actual Data 14
Regression Analysis Example Week x (Price) y (Appetizers) 1$ An analyst for a chain of seafood restaurants is interested in forecasting the number of crab cake appetizers sold each week. He believes that the number sold has a linear relationship to the price and uses linear regression to determine if this is the case.
Regression Analysis Example (cont.) Regression Statistics Multiple R0.843 R-Square0.711 Adjusted R-Square0.639 Standard Error Observations6 ANOVA dfSSMSFSignificance F Regression Residual Total CoefficientsStandard Errort StatP-value Intercept Price ($)
Least Squares Regression Line Appetizer Example 17
Interpretation of the Regression Intercept 18
Another Regression Analysis Example HoursScore A professor is interested in determining whether average study hours per week is a good predictor of test scores. The results of her study are: A student says: "Professor, what can I do to get a B or better on the next test. The professor asks, "On average, how many hours do you spend studying for this course per week?" The student responds, "About 2 hours." Use linear regression to forecast the student's test score.
Another Regression Analysis Example (cont.) 20 Regression Statistics Multiple R0.391 R-Square0.153 Adjusted R-Square Standard Error Observations8 ANOVA dfSSMSFSignificance F Regression Residual Total71300 CoefficientsStandard Errort StatP-value Intercept Study hours
Forecast Error Measures Bias Average error Variability Mean squared error (MSE) Standard deviation (s) Mean absolute error (MAD) Mean percent absolute error (MAPE) Relative bias Tracking signal (TS) 21
Summarizing Forecast Accuracy PeriodActual (A)Forecast (F)Error (E=A-F)Abs ErrorError Sq [(Abs E)/A] x Total MAD =23.9 MSE = s =34.8 MAPE =23.8% 22
Tracking and Analyzing Forecast Errors PeriodActual (A)Forecast (F)Error (E=A-F)Assessing bias: Cumulative forecast error (periods 1-9) = MAD (periods 1-9) = Tracking signal (periods 1-9) = Cumulative forecast error (periods 1-18) = MAD (periods 1-18) = Tracking signal (periods 1-18) = Assessing error variability/size: Total4 Standard deviation (periods 1-9) = 2s control limits for errors: 0 +/- 2(34.8) = / s Control Chart for Errors UCL = 69.6 LCL = -69.6
Forecast Performance of Various Forecasting Methods for a Medical Clinic Method Cumulative Sum of Forecast Errors (CFE – bias) Mean Absolute Deviation (MAD - variability) Simple moving average Three-week (n = 3) Six-week (n = 6) period weighted moving average w = 0.70, 0.20, Exponential smoothing = =