Let’s get perspectives from other modelers. It’s a fact that most models in scientific research papers are broken.
We need to be cautious of “Lies, damned lies, and statistics“, especially from data mining or machine learning. I speak from 20 years of financial markets modeling experience.
THIS!!!!!
I've been modeling and trading markets algorithmically for the last 10 years as well fellow pede. It is hard enough predicting a stock price 50 miliseconds into the future, yet these models predictinig well into the future were presented to the public as damn near certainties. imbalanced data sets, unknown variables, curse of dimensionality, chaotic modellng. We got terrible frequentist statics presented as fact. This is the problem with academic modeling and statistics- they are frequentist statistics! They do not work in the real world with vast amounts of unknowns. Had the models been bayesian in nature with a prior assumption that this 'pandemic' would be similar to all other pandemics we had seen, and then updated our assumptions as new data arrived we would not be in nearly as much of a panic and over-reaction.