08-07-2019, 07:51 PM
So again the TL;DR version is that the foundation of climate alarmism is based not on science, but data mining. Such models do provide ideas for further research, but are garbage for forecasting and policy choices.
Here's an article from Curry talking about the "scientific method/process" behind the models. Again, 100% spot on. Two of the tenets of good science, aside from a validated model, is that your methods and assumptions are documented and the work is reproducible (which obviously fails without the first, and when models are treated like proprietary trading models).
https://judithcurry.com/2010/10/10/the-culture-of-building-confidence-in-climate-models/
Here's a bit more on model fitting and validation just to prove I'm not pulling this out of my ass.
https://people.duke.edu/~rnau/three.htm
https://dziganto.github.io/cross-validation/data%20science/machine%20learning/model%20tuning/python/Model-Tuning-with-Validation-and-Cross-Validation/
LOL- "We learned that training a model on all the available data and then testing on that very same data is an awful way to build models" (a.k.a more derisively referred to as junk science).
And a link (it will download a pdf) to the IPCC chapter on model "evaluation" (there they go again choosing a similar sounding word but directly avoiding "validation"). Seems mainly devoted to providing a false sense of confidence in the models. That many things are tightly constrained still leaves plenty of giant assumptions and freedom fit a model. Actually an interesting read, and I think there's a pretty obvious disconnect between how and what is reported in the media and the range of uncertainty outlined here.
https://www.ipcc.ch/site/assets/uploads/2018/02/WG1AR5_Chapter09_FINAL.pdf
Here's an article from Curry talking about the "scientific method/process" behind the models. Again, 100% spot on. Two of the tenets of good science, aside from a validated model, is that your methods and assumptions are documented and the work is reproducible (which obviously fails without the first, and when models are treated like proprietary trading models).
https://judithcurry.com/2010/10/10/the-culture-of-building-confidence-in-climate-models/
Here's a bit more on model fitting and validation just to prove I'm not pulling this out of my ass.
https://people.duke.edu/~rnau/three.htm
https://dziganto.github.io/cross-validation/data%20science/machine%20learning/model%20tuning/python/Model-Tuning-with-Validation-and-Cross-Validation/
LOL- "We learned that training a model on all the available data and then testing on that very same data is an awful way to build models" (a.k.a more derisively referred to as junk science).
And a link (it will download a pdf) to the IPCC chapter on model "evaluation" (there they go again choosing a similar sounding word but directly avoiding "validation"). Seems mainly devoted to providing a false sense of confidence in the models. That many things are tightly constrained still leaves plenty of giant assumptions and freedom fit a model. Actually an interesting read, and I think there's a pretty obvious disconnect between how and what is reported in the media and the range of uncertainty outlined here.
https://www.ipcc.ch/site/assets/uploads/2018/02/WG1AR5_Chapter09_FINAL.pdf
--------------------------------------------------------
