Relations and Perspectives #4
Description
ML and DL (Tensorflow, Pytorch, Chain)
Their problems are hard and massive but well-defined with big data and specific objective metrics
Using ML and DL is to apply a more powerful weapon to solve the problems more efficiently and accurately
Mathematical Modeling
The problems are real-world, but usually very vague (even more real-world), you need to define the problem with rigor before exploring solutions
The problems can arise from everywhere from anything, it is more real-life oriented
you are allowed to use any tools and techniques to approach and solve them (use easy tools to train thinking process )
Java, Python, R
Great programming languages to build brilliant softwares
Great programming languages to processing and mining big data with clever algorithms
ABM on economics
Agent based modelling approaches to study and research economics, banking and money, trade and so on
What are more interesting and potentially demanding?
Mathematical modeling and ABM
- solve real world problem with simpler tools and languages (NetLogo, Extentions with python or R, Weka)
- time and energy focus on approaches, processes, defining and solving problems, rather than sharpening tools like Java, Python, Tensorflow and Pytorch
- Dynamics of Economies, markets, Money and Banking, Debt cycles, Trade wars are much more interesting and useful to policy making
- together with big real economic data, the model can be more realistic and dynamic to help policy makers
Simple toolkit is powerful enough
NetLogo, R, Weka are powerful enough to solve most of the problems probably
Learning to use them efficiently on real problems should be a demanding skill (also demand time and effort, and few people give to them)