Mathematical model for implementing Non Linear Time Series Forecasting using Deep Learning. Issue 2 (December 2020)

• Record Type: Journal Article
• Authors: Qureshi, Imran; Mohammad, Burhanuddin; Habeeb, Mohammed Abdul; Shaik, Mohammed Ali
• Is Part Of: IOP conference series. Volume 981:Issue 2(2020)
• Publication Date: 2020-12
• Subjects: mathematical model -- machine learning -- forecasting -- statistics -- time-series; Materials science -- Periodicals; 620.1105
• DOI: 10.1088/1757-899X/981/2/022021
• Languages: English
• ISSNs: 1757-8981

Abstract

Abstract: In this paper we propose a novel algorithm for implementing the mathematical model using the Non Linear Time Series Forecasting using the methods of deep learning which has to avoid noise and tends to transforms the space to next level in a iterative manner by swapping the down-trend to up-trend or up-trend to down-trend based on the iterative set values by computing the transformations. The proposed model must handle huge dimensional data which is in a dynamic nature of proposed model and then we need to compare the results attained from the proposed model with the complex ARIMA model and identified various trial and error methods for predicting the future price without guidance. For obtaining the results we used the petroleum dataset of India and attained the results which prove that our proposed mathematical model is far better than the famous ARIMA model while handling non linear time series applications.