Machine Learning Ex2 - Benchmarks

In my previous post, I implemented the algorithm for linear regression using gradient descent in Scala using two different methods: standard builtin mathematical methods and Scalala, a Scala linear algebra library.

Shortly after writing the solution I started to wondering if using Scalala had any performance impact on the runtime cost of the solution. While Scalala does have the overhead of object creation, it also makes heavy use of specialized classes, which should provide a considerable improvement.

I decided to do some naive benchmarking. These benchmarks are nowhere near scientific, but should provide a general sense of the solution's runtime. Since I was benchmarking the two Scala solutions, I decided to look at also the MATLAB/Octave and R solutions.

Machine Learning Ex2 - Linear Regression

Implementing linear regression using gradient descent in Scala based on Andrew Ng's machine learning course.