franfram / map-of-modeling

An attempt on making a map of the modeling world

Map-of-ModelingWorld

An attempt on making a map of the modeling world

inspired by https://www.youtube.com/watch?v=SzJ46YA_RaA

Why build a model? inference in science

Empirical models, Explanatory models

Bayesian Statistics Frequentist Statistics

Supervised Unsupervised Self-supervised Reinforcement Learning

No causes in, no causes out // No causes in, nothing out --> Causal modeling --> whisper (or use new better fair model) Richard's talk

ABC

When it's easier to describe change --> Differential equations (check 3Blue1Brown vids)

Symbolic regression

Linear models Non-linear models

DiffEq models

Models as information compression --> Check Marcus Hutter talk with Lex Fridman, whisper it (or use better model by fair)

Models as representations

Speaking math? check that paper by paul e smaldino

Approximation problems

Observation is imperfect --> Observation models

Causal models vs Statistical models

Top-down vs Bottom-up modeling.

ABMs vs (Multi-agent) Reinforcement Learning.

Computation in models

That (equation) is nice, but we can't calculate it --> Numerical computing

Algorithms used in modeling: MCMC,

Write with bookdown? with nvdev?

Other stuff to write about: How to (almost) screw your PhD in the attempt to make good science.

Forecasting -> specifying the dynamics vs learning them (NNs or symbolic regression)

Make a map of the modeling world and make a minimap of each section of the book

On the perks of being f**ked up --> no perks \(~)/

Title: A map of the modeling world a 100 page attempt to understand the modeling world.