A Study of the Indian Metal Sector Using Time Series Decomposition-Based Approach

Prediction of stock prices using econometrics and machine learning based approaches poses significant challenges to the research community since the movement of stock prices are essentially random in its nature. With the rapid development and evolution of algorithms using sophisticated machine learning and artificial intelligence concepts, of late, it has become possible to handle large volume of complex time series data for statistical analysis in real time using high-performance hardware and parallel computing architecture. Accordingly, robust predictive models are being built for achieving more accuracy in forecasting of highly random and stochastic stock price movements. This chapter presents a highly reliable and accurate forecasting framework for predicting the time series index values of the metal sector in India. First, a time series decomposition approach is followed to understand the behavior of the metal sector time series for the period January 2010 till December 2016. Based on the structural analysis of the time series, six methods of forecasting are designed. These methods are applied to predict the time series index values for the months of 2016. Extensive results are presented to demonstrate the effectiveness of the proposed decomposition approaches of time series and the efficiency of the six forecasting methods