Linear Regression in Rust

A simple linear regression implemented in Rust - because why not! the meat of this is the struct LinearRegression, representing a simple linear model.

  struct LinearRegression {
      betas: Vec<f64>,
      bias: f64,
      x: Vec<Vec<f64>>,
      y: Vec<f64>
  }

Then we have several implemented methods, new, step, fit, and predict which encapsulate the expected behavior. step uses a probably incorrect version of gradient descent.