Conceptual Overview: Explore how using different weights for averaging prior observations in a time series affects the forecast and the accuracy statistics.

Averaging prior observations is often a good way to forecast future observations for relatively stable time series. In the graph, the red dots with solid lines represent the time series data and the blue dots with dotted lines represent the moving average forecast.

Use the buttons at the bottom to select different moving average methods for the prior three observations. The "Naive Forecast" uses only the immediate prior observation. The "Moving Average" gives equal weight to the three prior observations. The "Weighted Moving Average" weights observations by recency. Compare how the different weighting systems perform as forecasts, both visually and in terms of the statistics, MAE (mean absolute error), MSE (mean squared error), and MAPE (mean absolute percentage error).

2 25 20 15 Sales (1000s of gallons) Naive Forecast: Wts {0,0,1} 10 MAE = 3.89 MSE = 17.67 MAPE = 20.23% 5- 9 10 11 12 0 1 2. 3 4 5 6 7 8 Week Naive Forecast Moving Average Weighted Moving Average 1. Out of the three methods, "Naive Forecast", "Moving Average", and "Weighted Moving Average", which produces a smaller mean absolute error? a. Naive Forecast b. Moving Average c. Weighted Moving Average d. None of the above a 2.Which method produces the smallest mean squared error? a. Naive Forecast b. Moving Average c. Weighted Moving Average d. None of the above