Selected article for: "early forecast and forecasting performance"

Author: Michal Ben-Nun; Pete Riley; James Turtle; David P. Bacon; Steven Riley
Title: National and Regional Influenza-Like-Illness Forecasts for the USA
  • Document date: 2018_4_27
  • ID: cheiabv0_54
    Snippet: In submitting forecasts based on uninformed mechanistic priors using an uncoupled 536 model at the start of the season, we failed to learn lessons that have been present in the 537 influenza forecasting literature for some time [30] . Historical variance is low during the 538 start of the season and the growth pattern is not exponential. Therefore, it would be 539 reasonable to forecast early exponential growth only in the most exceptional of not.....
    Document: In submitting forecasts based on uninformed mechanistic priors using an uncoupled 536 model at the start of the season, we failed to learn lessons that have been present in the 537 influenza forecasting literature for some time [30] . Historical variance is low during the 538 start of the season and the growth pattern is not exponential. Therefore, it would be 539 reasonable to forecast early exponential growth only in the most exceptional of not-yet-seen time points, are likely to perform better. Also, forecasting competitions 543 may want to weight performance differentially across time, with greater weight given to 544 forecasts during periods where there is a higher variance in incidence.

    Search related documents:
    Co phrase search for related documents