(noun)
parameters and/or random noise added to a predictive model to enable the model to appear to include a wider variety of outcomes than the fundamental mechanism of the model actually includes
(verb)
to add spurious adjustable parameters and/or random noise to a predictive model to enable the model to appear to include a wider variety of outcomes than the fundamental mechanism of the model actually includes
to artificially inflate your estimate of a very unlikely event occurring to ~30% with no justification within your model, simply so you can cover your ass should the unlikely event occur, all while retaining the option to claim that you predicted, with better than 2/3rd probability, that the unlikely event would not occur.
parameters and/or random noise added to a predictive model to enable the model to appear to include a wider variety of outcomes than the fundamental mechanism of the model actually includes
(verb)
to add spurious adjustable parameters and/or random noise to a predictive model to enable the model to appear to include a wider variety of outcomes than the fundamental mechanism of the model actually includes
to artificially inflate your estimate of a very unlikely event occurring to ~30% with no justification within your model, simply so you can cover your ass should the unlikely event occur, all while retaining the option to claim that you predicted, with better than 2/3rd probability, that the unlikely event would not occur.
“After realizing her model to predict economic growth didn’t include enough details, and was therefore predicting too narrow a range of possible outcomes, she nate-silver-thirty (ed) her model to get a more satisfactory appearing distribution of outcomes.”
by stats_definer August 28, 2020