Linear Regression
Linear regression is hardly used in practice for problems with huge datasets or problems that show complex relationships between the features and the targets. So what is the purpose of studying linear regression when neural networks produce much better results? Linear regression is a precursor to neural networks.
Info
Linear regression models can be interpreted as very simple neurons in a neural network.
Linear regression models are essentially the simplest neurons we can think of, therefore it is convenient to study the basic concepts that are used in all neural networks by studying linear regression. When we finally arrive at more complex deep neural networks, we will be familiar with the basic building blocks of neurons.
Warning
It is ok if at this point you do not yet understand the relationship between a neural network and the linear regression. Most likely you will need to go through the next three chapters to fully grasp the concept. Return here after you have finished the neural networks chapter to see if our definition of linear regression as a neuron makes sense to you.