Predictive analytics of chemical material pricing

• Main Author: Saini, Aditi
• Other Authors: Zhang, Jie
• Format: Final Year Project
• Language: English
• Published: Nanyang Technological University 2021
• Subjects: Engineering::Computer science and engineering::Computing methodologies::Artificial intelligence
• Online Access: https://hdl.handle.net/10356/147902
• Institution: Nanyang Technological University

Predictive analytics is the use of data, mathematical models and statistical algorithms to make predictions of the likelihood of future events. With the recent trend in technology and the ever-growing database of information, predictive analytics has become a catalyst to drive strategic decision making in most businesses today. This project explores the prediction of the prices of different raw material using time series- based analysis. Each raw material is composed of different chemicals, referred to as feedstocks. We investigate univariate and multivariate time series modelling techniques to create a generalised prediction pipeline using autocorrelation and cross-correlation analysis. We show through experiments and with the help of metrics (i.e., Mean Absolute Error and Mean Absolute Prediction Error) that how a statistical machine learning model outperforms the existing rule-based prediction model by identifying the underlying trends through correlation and time lag analysis.