Download
1 / 2
20 likes | 151 Vues
Initial predictions often fall short when compared to empirical models, but every innovation must start somewhere. Just as terrestrial weather forecasts were once ridiculed but have since matured, predictive modeling is evolving. This process of modelers providing forecasts for validation offers several advantages, including eliminating the need for institutions to adapt to new models and avoiding strict validation standards beforehand. It creates a "blind study" environment while freeing modelers from handling validation data. Active involvement and funding are crucial for ongoing improvements.
E N D
Why? • Initial predictions will be lousy and probably worse than empirical models. • BUT, one needs to start somewhere. Terrestrial weather predictions were laughable when they started but have now reached maturity. From: Kalnay, 2003
Validation • The process of modelers providing the forecasts for validation has a number of advantages: • No need for other institutions to install and “learn” the model. • No need to adhere to strict standards before model is validated (pitfall in traditional approach). • Provides “blind study” for modelers who do not know who might scrutinize their output. • No need for the modelers to deal with validation data. • BUT, modelers need to be involved funding, in order to get here: From: Kalnay, 2003
