Background Knowledge:

Model fitting is a problem that has been around as long as physics. The main purpose of physics is to reliably predict the changes in the world around us. Unfortunately while research has been done in this field, it is quite limited and we can only fit models to specific datasets. While this is the case, we have simpler methods for the time being, such as neural networks.

Quick Run:

The first step to model fitting is to find the variables we wish to vary. This task can take quite a long time, as we see when we look at the science of physics. Once we determine the variables, we can start plotting them on a graph. This only really works when there is 1-2 very well defined variables, more than this and we run into complexity issues. Thankfully other forms of computer science came up with a solution to this. Rather than make a precise model fit for an object, make a bounding object that adheres to a certain model. Next match the bounding object to the dataset and if there is a match of a certain threshold or higher, we identify the object as the bound object. This is a very common technique and was used quite a lot in my Computer Vision experience. A perfect example here is the superquadratics which i talk about in another post.

Most Important Tidbits:

As of this time I cannot think of any really interesting tidbits to add here.