Predicting NYC Lot Prices with Lathe in Julia

That’s right ladies and gentlemen, introducing: The first Lathe.jl model, and unsurprisingly; it’s Linear Regression.

Julia is a fantastic, powerful language (especially if you like programming arduino boards.) With that being said, given that Julia is very much a functional language , its almost in-genius Base structure make it very usable for what would traditionally be considered object oriented programming. However, a target audience of Julia that is relatively lacking is Data-Scientists.

Although a-lot of Julia’s veterans may not have a problem with Pkg.add()-ing seven different packages for 3 different models, this turned out to be my biggest gripe with Julia. Julia has an absolutely fantastic (not to mention easy) package, Flux.jl for gradients, image-recognition, all that fun stuff, but unfortunately I feel that packages come up short for predictive machine-learning, especially unsupervised learning. With my plans to change this, by putting a-lot of data-science tools in one place for Julia Data-Scientists (including myself) I ended up starting a small package, at first it was a few scalars, t test, f test, and at a certain point I moved on to thinking of creating my own model module, which has occurred over the past 2 months.

Although I don’t think that my package is life-changing (just yet,) I think that even in its pre-release state, I still opt to use it over the more traditional options Julia has to offer (mostly at the benefit of my local storage.) So today, I am proud to present to you Lathe.jl’s official SimpleLinearRegression.