Analysis of Prediction Techniques for Temporal Data Based on Nonlinear Regression Model

Abstract

This paper depicts algorithms for the prediction of frequent data item that is: frequent item prediction method (FIPM) and frequent temporal pattern data stream (FTP-DS) based on linear and nonlinear regression algorithms, which are used to predict the trends in sequence with stream data. Sliding window is used for gathering and preprocessing real-time stream data. FIPM and FTPDS algorithms compute Support(y) for appointed sequence and describe linear and nonlinear equations to forecast sequence trends in the future. In this paper, the system used linear and nonlinear regression-based algorithms to evaluate the mean square error (MSE) during prediction.