Model vs algorithm

In general:

  • a model takes an input x and maps it to an output y
  • an algorithm is a method or a process that can be defined with certain steps, e.g. like gradient descent. The algorithm can terminate when certain quality criteria is reached, e.g. when an optimum is reached resp. a condition is fulfilled.

My take on that particular question:

  • So a linear regression w/ y=f(x) = m x + b is a statistical model since it transforms the input x linearly to the model output y
  • More specifically: the way you fit the model when the models parameters are learned can be described with an algorithm, e.g . Gradient descent in the context of linear regression:


In my opinion it depends on the context here on what you want to address:

  • the fitting process (then the term algorithm is totally fine)
  • or if you mean the (parametrized) model that maps an input to an output. Then you better call it just a regression model.

Hope that helps, @mvrbiguv! Please let me know if anything is unclear.

Side note: if you want to check out how linear regression can be solved with the normal equation, this thread might be worth a look.

Best regards