Video 2.2 ///Short Note
· Nearest neighbor averaging has smoother version such as kernel ad spline smoother later in the course.
· The first neighborhood has the 10% of data points. When we talk about the 2D it is less than the 1D.
· When we are going to 5D OR 10D its really hard to identify the neighborhood.
· The liner model is an important example of parametric model.
· We can see that in to fit the data it is needed to have x2.
· It I more important to have the data with linear regression model to have the data for the plotting the graphs.
· Even the overfitting the training data there may be many errors.
· Linear models are easy to interpret.
· We can have good fit, over fit or under fit.
· We often prefer a simple model involving fewer variables over a black-box predictor involving them all.