Agent-MB-DivClues: Multi Agent Mean based Divisive Clustering

Abstract

The conventional way of performing data search comprises of disadvantages specifically in environment induced and non-stationarity while performing advanced prediction. In this context we propose an agent based MBDivclues hierarchical clustering method which is based on Multi Agent System (MAS) approach which enables us with decentralized control over the framework to attain the similarity among objects to structure various categorical attributes and measuring of relative distance among various numeric values by calculating the arithmetic mean which is assigned to the rootnode. In this paper we propose a novel divisive hierarchy that is based on the construction and implementation of a non-binary tree using multi agents. And to predict the forecasting error and be a step ahead In this approach we use a system of several adaptive forecasting agents, where the novel multi-agent forecasting system allows an agent to perform minimal training sequences and to attain accurate results to forecast by employing single prediction algorithm.