Yousef Karsh had some kind of connection to Boston’s Brigham & Women’s Hospital (I forget what the connection was), and there are several of his large-format photos, including the Winston Churchill, in the library adjacent to the Family Center there. Seeing ol’ Winston there, 1.5 times life size, staring down at you is quite the experience.
Ai-generated code is known to be buggy, at least sometimes, so I assume you'll compare your Julia results to R results given the same data? Be very interesting to see if you get identical results.
Back when I was still working, a manager I knew tried to get me interested in R (because he was getting into it), but stats were never my thing. But it has sort of remained on the edge of my radar because of my interest in baseball stats. Python is also, as you say, "light in the loafers" when it comes to serious stats. (But I wonder if there isn't a Python package for stats or that can leverage an R install. Or, given how popular Python has become, both.)
Definitely can’t just give AI its head and let it gallop off wherever it wants. I use it primarily for the picky syntax details I’m not yet fluent in. The output of lm(@formula(mpg ~ hp +drat +wt, dp)) in Julia matches R on the same real world data set except for the loess lines. This is due to the Loess’s package implementation details varying from R, but is not material. RCall does provide facilities to run R code from within Julia. That comes, however, with a performance hit and added dependency maintenance overhead.
I hear ya on the dependency issues. I lived there when I was doing Windows programming in C++ and VB. It's also a problem for Python programmers using third-party packages. Definitely nice if you can keep it all in your home ballpark.
Yousef Karsh had some kind of connection to Boston’s Brigham & Women’s Hospital (I forget what the connection was), and there are several of his large-format photos, including the Winston Churchill, in the library adjacent to the Family Center there. Seeing ol’ Winston there, 1.5 times life size, staring down at you is quite the experience.
Ai-generated code is known to be buggy, at least sometimes, so I assume you'll compare your Julia results to R results given the same data? Be very interesting to see if you get identical results.
Back when I was still working, a manager I knew tried to get me interested in R (because he was getting into it), but stats were never my thing. But it has sort of remained on the edge of my radar because of my interest in baseball stats. Python is also, as you say, "light in the loafers" when it comes to serious stats. (But I wonder if there isn't a Python package for stats or that can leverage an R install. Or, given how popular Python has become, both.)
Definitely can’t just give AI its head and let it gallop off wherever it wants. I use it primarily for the picky syntax details I’m not yet fluent in. The output of lm(@formula(mpg ~ hp +drat +wt, dp)) in Julia matches R on the same real world data set except for the loess lines. This is due to the Loess’s package implementation details varying from R, but is not material. RCall does provide facilities to run R code from within Julia. That comes, however, with a performance hit and added dependency maintenance overhead.
I hear ya on the dependency issues. I lived there when I was doing Windows programming in C++ and VB. It's also a problem for Python programmers using third-party packages. Definitely nice if you can keep it all in your home ballpark.