Big Data Analytics - Industry Perspective I

The purpose of this session is to highlight industrial successes and challenges of Big Data Analytics. Presentations will cover how process industries (such as chemical, oil & gas, semiconductor, energy, pharmaceutical, and food) find actionable insights from Big Data to improve product quality, improve productivity and yield, and save money using this step: Data -> Information -> Knowledge -> Wisdom Analytics based on data-driven models (such as multivariate analysis, machine learning & data mining techniques) are typically used in process industries. Presentations will address the Big Data 4 Vs (Volume, Velocity, Variety, and Veracity): Volume: Typical data historian stores thousands of process variables every minute. Data volume is an important aspect to consider in off-line multivariate batch data analysis Velocity: Real-time process monitoring and control need to address process dynamics. On-line data are available rapidly and therefore model maintenance issues are of particular interest here Variety: Process industries collect data from various sources including raw material, process, product quality, customer feedback, and other unstructured sources (i.e., text data). Analytics are more insightful when data are coming from multiple sources Veracity: Data pre-processing approaches are used to address data accuracy and integrity issues (such as missing data, outlier, noisy data, and batch alignment)