Economics postgrad, I'm doing some stuff on algorithmic game theory, and I'm basically trying to model human behaviour.
So far I only know MATLAB and excel, but I want to make more sophisticated programs.
Anyways what would be the best/easiest language for me to transition into, because I'd be learning this in my spare time as it's not part of my course exactly.
Other urls found in this thread:
python
/thread
Python, Julia, or R
I recommend Julia.
+ the relevant scientific libs
This.
If you're doing a high level data analytics, Python has a lot of tasty modules you can use.
Scilab actually shares the same kernel as MATLAB.
For adhoc (interactive) data analysis, statistics and plotting R (Rstudio and tidyverse) is top tier.
For coding custom algorithms or using ML Python is good, as long as you don't need C-tier performance or can vectorize everything
>Scilab actually shares the same kernel as MATLAB.
No.
If you're doing economics, you might want to look into R. A lot of economics and stats people use R, and there is a strong enough community so that you can find pretty good toolboxes, examples, and support.
However, the syntax is very weird and very not-programming-like. So it takes time to get used to it.
You might also want to spend some time learning Business Intellingce tools such as Tableau or Power BI.
>the syntax is very weird and very not-programming-like
That true but I don't think most people should be "programming" in R. The point of R is that it has so many top-tier libraries (ggplot, data.table, etc.) that basically anything you need is a straightforward declarative one-liner.
If you have "control flow" in R you're probably doing it wrong. It's a good thing desu, since no one wants to debug their own complex code while they're trying to understand out some data
Yeah I know R, but I'm trying to construct learning behaviours about new conditions, simulating how people would react to new regulations and what not. Stuff like crowd behaviour, it's more than just statistics.
Like I can write the logical structures of the algorithms, but I need something to run it.
I guess the question is what do you know want to do now, that you cannot do with Matlab + excel?
Matlab is quite powerful and its limitations are only apparent when you try to build big OOP projects or go into complex programming patterns.
MATLAB is quite annoying when you're trying to keep tracking of who's doing what.
For example if I want to know who will most likely form alliances or choose to cooperate, I'd have to do some weird shenanigans with adding values onto variables anytime a person does this or that, to keep track of what's going on.
It works, but it isn't really all that elegant and I kinda want to see what I could do with other softwares.
this
As a Physicist I agree. Everywhere I go, Python is the "go to" language for data analysis, be it your run down shitty basement lab or CERN.
>CERN
C++ is still used more widely than pyROOT I think.
That depends on what kind of data you need to analyze... or rather, how much data you need to analyze compared to your computing power available. While places like CERN have a fuckton of money to buy computing power, many "public" institutes have limited resources where it can happen that 20-40 people work on a single server that hast 16 cores, 2GHz per core. In such a case, Python is the "go to" language usually.
However, if you have more computing power available, like a stand alone computer with 16 and more cores for a single experiment/measurement, you tend to use C++ yes. I'm talking about the "regular physics lab" here, not huge experiments like the LHC. There, you basically do needed computing power to cost analysis and go from there, the "fastest" language wins.
>Julia
Avid R user here but have always been curious about Julia. What's the benefit?
R can do that. Don't trust people who say you can't be used for "programming" or making control structures in R. The only reason people say that is because R is inefficient when compared to something like C but you will spend way less time writing code in R and in the end, your total time spent on the project will break even. In short, R is not suited toward creating software but anything in terms of algorithmic modeling or data analysis is fair game.
On CERN?
Yes, and there is also an insane amount of FORTRAN code that nobody understand how it works but works and is fast so nobody can replace it.