CISC483 Lecture Notes - Lecture 5: Data Mining, Dimensionality Reduction
14 Sep 2016
School
Department
Course
Professor
Document Summary
Different departments use different conventions, time periods, primary keys, etc. Overlay data data from outside an organization needed for data mining and which must be integrated with organization data. Aggregation within data have to determine how multiple potential entries should be aggregated to provide instances for analysis individual long distance calls vs total customer minutes to different states or total customer minutes. Dimensionality reduction more attributes that can be effectively dealt with need to reduce the # of attributes that are considered. Denormalization process of taking two or more relations and joining them together into a single relation (today"s notes) Aarf files standard way of representing machine learning data sets as flat files. {} a form of inductive learning given a collection of examples of the form (x, f(x)), return a function h that approx f h is the hypothesis greedy, divide-and-conquer approach. Each internal node corresponds to a test of the values of an attribute.
