IDEA—a conversational, heuristic program for Inductive Data Exploration and Analysis

Abstract

The goal of this project is to improve the power and scope of computer routines that search for structure in a data base. IDEA (Inductive Data Exploration and Analysis) is a computer program that detects and represents inherent structure in multi-variage data. The IDEA program is designed either to run in an automatic mode or to allow the investigator to intercede at each major decision in the analysis. At each such juncture he is presented with information that permits him to concur or override a computer decision before the program continues to the next major decision. In addition, IDEA has two other distinguishing features: (1) Heuristic computational procedures are used for those cases where the combinational aspects of the analysis would require extensive computations, and (2) heuristics are selectively used for different types of data, enabling IDEA to operate on a mixture of nominal (categorical), ordinal (ranked), and interval- or ratio-scaled measurements.