CAP 4770 - Lecture 3.2: Discretization and Binning

Flashcards covering key concepts of Discretization and Binning methods.

Discretization or Binning

The process of converting a continuous attribute into a discrete (categorical) attribute, often required for algorithms that only work with discrete data.

Equal-Width Discretization

A technique where the range of an attribute is divided into N bins of equal width. The bin width is calculated as (max - min) / N.

Bin Width Formula

bin_width = (max – min) / N, where 'max' is the maximum value of the attribute, 'min' is the minimum value, and 'N' is the desired number of bins.

Equal Frequency Discretization

A technique that ensures each bin holds the same number of values (same frequency of values), dividing the range of values so each bin contains an equal count of training examples.

Unsupervised Discretization

Discretization methods where the labels of the data (e.g., class labels for classification) are not taken into account during the binning process. Equal-Width and Equal-Frequency are examples.