We want to use this data to classify a new fruit, e.g. \(x = (195g, \text{yellow})\). What do you think, is it more likely to be an apple or a banana?
Now we consider non-parametric models for classification, i.e. in contrast to the previous approaches, we do not try to determine the parameters of a separating hyperplane or a neural network. Instead, we aim to use the data to approximate their distribution.
To motivate this, consider the training dataset
We want to use this data to classify a new fruit,
e.g. \(x = (195g, \text{yellow})\).
What do you think, is it more likely to be an apple or a banana?
Next up: 2.2.3.1 Bayes Classification