Analysis and implementation of data mining algorithms for deploying ID3, CHAID and naive bayes for random dataset

Abstract

Objectives: Effective data processing for fast retrieval of information has become a burning issue. A modern document contains not only text but images, video, audio as well. In this paper, a brief history of storage devices from the Vedic period to the world of digitization with some important inventions has been presented. Method/Statistical Analysis: It also includes the discussion on how data is transformed for the decision making process along with preprocessing techniques. Findings: A comparative analysis has been done of various techniques, their specific algorithms, uses, pros, limitations and applications where these can be implemented. It helps to give us an insight about these techniques. Finally experimental results on three different algorithms (Id3, CHAID, Naive Bayes) using Rapid Miner have been evaluated to compare their performance based on three parameters (accuracy, precision and recall). Applications/Improvements: The empirical results show the ID3 as more accurate than others with 95.95% accuracy while CHAID shows 89.11% and Naive Bayes classified 81.77% data accurately.