The simplest algorithms we can use for machine learning are linear models. In this video we talk about what makes a model linear and why this means more than just y=mx+b. We also explain nonlinear models with an example of materials data. We examine which is better by plotting residuals and looking for bias and total error.
Check out the whole materials informatics series at https://youtube.com/playlist?list=PLL0SWcFqypCl4lrzk1dMWwTUrzQZFt7y0 with workbooks and course notes available at https://github.com/sp8rks/MaterialsInformatics
0:00 defining and comparing linear vs non-linear models
3:44 example of linear and non-linear fits of material data
6:35 overfitting and variance vs bias in linear vs non-linear models
8:15 regularization to prevent overfitting (lasso vs ridge)
12:35 worked example of linear non-linear fitting in python
