摘要: |
This research note uses two Time Series techniques, Holt-Winters (HW) Algorithm and Autoregressive Moving Average Model (ARMA), to predict motor vehicle crash fatalities during six holiday periods. Based on the data from 1983 to 2001, the estimated fatalities in 2002 are: 564 and 564 (New Year), 501 and 514 (Memorial), 748 and 755 (4th of July), 486 and 498 (Labor), 572 and 565 (Thanksgiving), and 156 and 145 (Christmas). Incorporating the actual 2002 fatality counts in the data series, the forecasts for the 2003 holiday periods are: 141 and 141 (New Year), 508 and 504 (Memorial), 537 and 564 (4th of July), 526 and 545 (Labor), 547 and 564 (Thanksgiving), and 548 and 576 (Christmas). |