Computational Business Analytics.pdf
Subrata Das is the founder and president of Machine Analytics and also serves as a consulting chief scientist. He is an editorial board member of the Information Fusion journal. He earned his Ph.D. in computer science from Heriot-Watt University.
Background Introduction at analytics Probability and statistics basics Mathematical logic basics Introduction to uncertainty Theory of algorithmic complexity Performance measures Statistics for Descriptive and Predictive Analytics Descriptive statistics Inferential statistics Dependence methods Interdependence methods Potential enhancement and augmentation of statistical techniques Analytics Problem Modeling in Symbolic Artificial Intelligence Approaches to handling uncertainty Deductive, inductive, and abductive reasoning Ontology and knowledge reasoning Ontology and knowledge representation Rule-based system Bayesian belief networks (BBN) Case-based reasoning Machine Learning/Data Mining for Descriptive and Predictive Analytics Generative vs. discriminative models Supervised, unsupervised and semi-supervised learning Symbolic techniques Sub-symbolic techniques Algebraic Bragging and boosting Time-Series Modeling for Predictive Analytics ARMA/ARIMA ARCH/GARCH Hidden Markov models (HMM) Dynamic Bayesian networks (DBN) Kalman filtering and extensions Particle filtering Prescriptive Analytics and Decision Support Test hypothesis Expected utility theory (EUT) Influence diagrams Symbolic argumentation Reinforcement learning Markov decision process (MDP) and partially ordered MDP Text Analytics Natural language processing (NLP) Text classification Information extraction and representation in RDF Case Studies Text document classification Image classification Topic detection Customer segmentation Syndromic surveillance Clinical state estimation Risk assessment via text analytics Opinion mining and sentiment analysis Index References
Traditional business analytics have so far focused mostly on descriptive analyses of historical data using a myriad of sound statistical techniques. This book describes how numerical statistical techniques can be augmented and enriched with techniques from symbolic artificial intelligence (AI), machine learning (ML)/data mining, and control theory for enhanced descriptive, predictive, and prescriptive analytics. The book is unique in its coverage of both traditional probabilistic/statistical and cutting-edge AI/ML-based approaches to descriptive and predictive analytics and associated decision support. It provides analytics practitioners with problem modeling guidance and appropriate modeling techniques and algorithms suitable for solving practical problems. The book offers a detailed account of various types of uncertainties and techniques for handling them. Special emphasis is given to modeling problems that are time-dependent. The book also covers text analytics with useful applications, such as information structuring and sentiment analysis.