Credits: 3 (3-0-0)
Description
Process data pre-processing and handling: data visualization and transformation; quick revisit to regression modeling. Dimensionality reduction and latent variable models with applications to fault detection and inferential modeling of processes: Principal Component Analysis (PCA), factor analysis, canonical correlation analysis, partial least squares; Classification and clustering methods with applications to process mode diagnosis: k-nearest neighbor, naive Bayes, linear discriminants, support vector machines, decision trees and forests, k-means, fuzzy c-means, possibilistic c-means, hierarchical clustering methods, mixture models; Nonlinear approaches: Kernel methods- kernel PCA, kernel SVM, neural nets-feed forward networks, Gaussian process; Entropy and its applications to redundant variable isolation: Shannon entropy, cross entropy, joint and conditional entropy, KL-divergence, mutual information;Model learning approaches: Maximum likelihood, maximum a posteriori, Bayesian approaches; Expectation-Maximization, back propagation, ensemble learning; Model assessment and validation: BIC, kfold cross validation, model averaging; Switching process systems modeling: Markov models, hidden Markov models (HMM); Estimation and inference of dynamical systems: Kalman filter and smoother, particle filters; Introduction to software packages: PYTHON, MATLAB and R; Process applications and case studies: Continuous Stirred Tank Reactor example and Tennessee-Eastman process case study.