320 likes | 432 Views
HMP 654 Operations Research and Control Systems in Health Care Fall 2014. Forecasting - Introduction. Forecasting in Health Care Forecasting Models Structural Models Time Series Models Expert Judgment Time Series Models: Demand has exhibited some measurable structure in the past.
E N D
HMP 654Operations Research and Control Systems in Health CareFall 2014
Forecasting - Introduction • Forecasting in Health Care • Forecasting Models • Structural Models • Time Series Models • Expert Judgment • Time Series Models: • Demand has exhibited some measurable structure in the past. • The structure will continue into the future.
Forecasting - Time Series • Signal vs. Noise • Extrapolation Models • Accuracy of Forecasts
Forecasting - Stationary Models • Stationary Time-Series • Moving Averages
Forecasting - Moving Avgs. 33 + 38 2 38 + 31 2 33 + 38 + 31 + 35 4 SUMXMY2(B7:B26,D7:D26)/COUNT(D7:D26)
Forecasting - Weighted M.A. • Weighted Moving Averages
Forecasting - Weighted M.A. 0.3 x 33 + 0.7 x 38 0.3 x 38 + 0.7 x 31
Forecasting - Weighted M.A Finding the Optimal Weights
Forecasting - Weighted M.A. Finding the Optimal Weights MSE vs W2 W2
Forecasting - Weighted M.A. Finding the Optimal Weights
Forecasting - Weighted M.A. Finding the Optimal Weights
Forecasting - Exp. Smoothing • Exponential Smoothing
Forecasting - Exp. Smoothing 0.7 x 33 + 0.3 x 33 0.7 x 38 + 0.3 x 33
Forecasting - Holt’s Method • Compute the base level Et for time period t using equation 11.6 • Compute expected trend value Tt for time period t using equation 11.7 • Compute the final forecast Y^t+k for time period t+k using equation 11.5
Forecasting - Holt’s Method Initial base level = first demand value Set initial trend to 0 Forecast for Qtr. 3, 1990: 634.2= 0.5 x 584.1 + (1 - 0.5) x (684.2 + 0) -25 = 0.5 x (634.2 - 684.2) + (1 - 0.5) x 0 609.1 = 634.2 + 1 x (- 25)
Forecasting - Regression • Linear Trend Model
Forecasting - Regression • Linear Trend Model
Forecasting - Regression • Quadratic Trend Model
Forecasting - Regression • Quadratic Trend Model
Forecasting - Seasonality • Adjusting trend predictions with seasonal indices 102 + 107 + 106 + 108 + 106 5
Forecasting - Seasonality • Use of Seasonal Indices • Create a trend model and calculate the estimated value for each observation in the sample. • For each observation, calculate the ratio of the actual value to the predicted trend value • For each season, compute the average of the ratios calculated in step 2. These are the seasonal indices. • Multiply any forecast produced by the trend model by the appropriate seasonal index calculated in step 3.