Lathe.jl is an all-in-one package for predictive modeling in Julia. It comes packaged with a Stats Library, Preprocessing Tools, Distributions, Machine-Learning Models, and Model Validation. Lathe features easy object-oriented programming methodologies using Julia's dispatch.
(press ] in the Julia REPL to go to Pkg REPL)
(v1.3) pkg> add Lathe(v1.3) pkg> add Lathe#UnstableLathe brings an entirely different methodology to the Julia language. Types are created to adhere to the object-oriented programming paradigm, and syntax is akin to that of Pythonic machine-learning packages, like SkLearn.
Lathe includes many of the tools commonly used by machine-learning engineers and scientists out of the box, rather than relying on more dependencies to do so.
Lathe models can be easily serialized and deployed onto production servers using Genie.jl, or a similar high-performance web-server. Lathe also has support for pipelines, meaning most pre-processing operations can be automated and performed with one easy call.
Lathe uses a faster methodology than most other Julia packages for machine-learning. Furthermore, the package also takes advantage of the natural ability of the language to be fast. As a result, Lathe is also faster than most similar packages for other high-level statistical programming languages.
Lathe is written in 100-percent pure Julia. As a result, the package often takes advantage of very Julian methods of dealing with problems, such as dispatch, macros, and syntactical expressions.
Lathe.jl has DataFrames.jl support, making it incredibly easy to put your data to work!
Lathe.jl is a relatively immature package, but is still making considerable leaps in the Julia ecosystem! Lathe is constantly improving, and that is a great thing if you happen to be utilizing the package!
