ISE 789 / OR 791: OPTIMIZATION MODELS FOR SYSTEMS ANALYTICS - A Graduate Level Special Topics Course Shu-Cherng Fang North Carolina State ...
←
→
Page content transcription
If your browser does not render page correctly, please read the page content below
ISE 789 / OR 791: OPTIMIZATION MODELS FOR SYSTEMS ANALYTICS A Graduate Level Special Topics Course Shu-Cherng Fang North Carolina State University – Spring 2022
Systems Analytics
• Modern information technology (IT) provides industries
with the ability to collect vast amounts of data (Big Data)
of underlying business systems.
• Data requires more than just flashy dashboards and
reports. Information and knowledge (Data Mining and
Machine Learning) are the key.
• Systems analytics help us make better data-driven
business decisions for industries (Analytic DM).
•Operations Research - SAO • OR models can be useful for machine learning and analytic decision making with dynamic or transactional data (AI). • SAO provides mathematical/analytic theory and tools for modeling system problems of interests to make optimal decision making using dynamic big data.
New initiatives of SAO area • A sequence of four courses on systems analytics 1. Optimization models for systems analytics 2. Stochastic models for systems analytics 3. Statistical models for systems analytics 4. Simulation models for systems analytics
OPTIMIZATION MODELS
FOR
SYSTEMS ANALYTICSOutline • Course objective • Pre-requisites • Course content • Grading and exams • Classroom rules • Textbook and references
Course objective • This graduate-level course intends to provide the fundamental theory and solution methods of commonly used optimization models for analytic system decision making with dynamic data. • It involves mathematical modeling, analytic methods, machine learning, and data-driven business decision making. • It is a flexibly structured experimental course that will evolve to a regular course.
Important to know • This special topics course is designed mainly for PhD students with proper background, research interests and self-learning capabilities. • This course does NOT teach you how to concretely model a system problem. • This course does NOT tell you the details of all theory and algorithms of mathematical optimization. • This course does NOT tell you how to manage existing data banks. • This course does NOT tell you how to use Excel, SAS OPT, MATLAB, Python, LINGO, CPLEX, CVX, or any commercial solver and software platform.
Important to know • This course INTENDS to prepare you for understanding the basic optimization models and skills involved in systems analytics. • This course INTENDS to prepare you for appreciating more in reading existing literatures. • This course INTENDS to prepare you for conducting research in the area.
Prerequisites 1. ISE/OR/MA 505: Linear Programming 2. Programming using CPlex, Gurobi, CVX, SeDuMi on MATLAB or equivalent (OR/MA 706: Nonlinear Programming or ISE/OR 708: Integer Programming will help.)
Course contents
I. Optimization models
(i) Linear Optimization
(ii) Integer Linear Optimization
(iii) Quadratic Optimization
(iv) Linear Conic Optimization
(v) Nonlinear OptimizationCourse contents
II. Machine learning approaches
(i) Supervised learning for classification and prediction
Support Vector Machines & Regression (SVM & SVR)
Artificial Neural Networks (ANN)
(ii) Unsupervised learning for clustering and featuring
Similarity Learning and Sparse Solutions
(iii) Reinforcement learning in dynamic environment
Markov Decision Process and Dynamic Programming
(789-II Stochastic Models for Systems Analytics –
Prof. Yunan Liu in Spring 2021)Course contents
III. Systems and data
(i) Knowledge systems
- collaborative filtering (Jester, MovieLens datasets)
- classification (UCI repository)
(ii) Financial investment systems
- Markowitz mean-variance portfolio selection
(S&P’s Compustat North America database)
(iii) Electric systems
- data compression and sparse signal recovery
(Sparco database)
(iv) Power systems
- load forecasting (GEFCom2012)
- unit commitment (MATPOWER)
(v) Aviation systems
- failure prediction (NASA aircraft engine dataset)Machine learning and optimization
1. Many machine learning problems are formulated as minimization of
some loss function that measures discrepancy between the
predictions of the model being trained and the actual problem
instances, or as maximization of some reward function that affirms
an expected decision.
2. One major difference between machine learning and optimization
lies in the goal of generalization - optimization intends to minimize
the loss/maximize the reward on a set of seen examples while
machine learning is concerned with minimizing the loss/maximizing
the reward on unseen samples.Some thoughts on artificial intelligence 1. The main goal of artificial intelligence (AI) is to equip machines (computers) with human (or super-human, hopefully) intelligence for dynamic decision making. 2. Computers examine data to extract embedded information (data mining) to form useful knowledge (machine leaning) for right decision making (analytics). 3. Algorithms guide machines to perform each desired task step by step.
Grading and standard 1. Homework - 50% 2. Course project - 30% 3. Exam - 20% A - 88 and above B - 75 to 87 C - 60 to 74 Fail - under 60
Homework / project / exam • Homework (5) - biweekly • Project (1) - end-of-semester report and presentation - at most 2 persons a team • Exam (1) - after finishing Part II - comprehensive, take-home
Classroom rules • Rule 1: No late homework without pre- approval. • Rule 2: Turn in your homework through email. • Rule 3: Office hour can be held through Google Meet. • Rule 4: No make-up exam without pre- approval or an official "doctor’s note".
References
1. S.-C. Fang and S. Puthenpura, "Linear Programming
and Extensions,” 1993, Prentice Hall,
ISBN 0-13-037581-0
2. D. G. Luenberger and Y. Ye, "Linear and Nonlinear
Programming,” 4th Edition, 2016, Springer,
ISBN 978-3-319-18842-3 (eBook)
3. S. Boyd and L. Vandenberghe, “Convex Optimization,”
7th Printing, 2009, Cambridge University Press,
ISBN 0-521-83378-7
4. L. A. Wolsey, “Integer Programming,” 1998,
John Wiley & Sons, ISBN 0-471-28366-5References 5. S.-C. Fang and W. Xing, “Linear Conic Optimization,” Lecture notes (in Engllish), book (in Chinese), 2013, Science Press, ISBN 978-7-03-038176-7 6. E. G. Birgin and J. M. Martinez, “Practical Augmented Lagrangian Methods for Consttrained Optimization,” 2014, Society of Industrial and Applied Mathematics, ISBN 978-1-611973-35-8 7. P. Flach, “Machine Learning,” 2012, Cambridge University Press, ISBN 978-0-511973-00-0 8. T. Hastie, R. Tibshirani and J. Friedman, “The Elements of Statistical Learning,” 2nd Edition, 2009, Springer, ISBN 978-0-387-84857-0
Available Software (Packages)
• Support Vector Machine (SVM)
▪ Matlab
▪ Install instruction: https://software.ncsu.edu/supplied_software/matlab/
▪ Tutorial:
o https://www.mathworks.com/help/stats/support-vector-machines-for-binary-
classification.html
o Vedio: https://www.youtube.com/watch?v=q778MSq21vU
o Tips: try to type ‘help fitcsvm’ in the Command Window of your Matlab.
o Advanced library: LIBSVM https://www.csie.ntu.edu.tw/~cjlin/libsvm/
▪ Python
▪ Tutorial:
o “Sklearn’’ library (most popular):
https://scikit-learn.org/stable/modules/svm.html
https://www.kaggle.com/prashant111/svm-classifier-tutorialAvailable Software (Packages)
• Neural Networks
• Matlab
o Getting started with Neural Networks using MATLAB
https://www.mathworks.com/videos/getting-started-with-neural-networks-
using-matlab-1591081815576.html
o Video: https://www.youtube.com/watch?v=WoLBeWclYxo
o Advanced documentation: Neural Network Toolbox User's Guide –
Description (google it)
• Python
o ‘‘Keras’’ Library:
https://machinelearningmastery.com/tutorial-first-neural-network-python-keras/
Examples: https://keras.io/examples/
o “TensorFlow” library:
https://www.tensorflow.org/tutorials/quickstart/beginner
https://machinelearningmastery.com/tensorflow-tutorial-deep-learning-with-tf-
keras/You can also read
