BUS 6131 - Predictive Analytics

3 lecture hours 0 lab hours 3 credits

• Evaluate, propose, and apply appropriate predictive analytics techniques for a given decision-making scenario
• Illustrate the benefits of business analytics for individuals, stakeholders, and organizations
• Develop and deliver clear and persuasive messages based on the outcomes of analytics projects
• Build decision models using supervised and unsupervised learning algorithms
• Implement predictive models using supervised learning methods of linear, polynomial and logistic regression, support vector machines, decision trees and forests, and gradient boosting trees and artificial neural networks
• Implement predictive models using unsupervised learning methods of clustering, k-means analysis, hierarchical clustering, anomaly detection and reinforcement learning

• Python language environments
  • I-Python (Interactive Python) console mode, Integrated Development Environment (IDE) and Jupyter Notebook environment.
  • Installation, configuration, and maintenance of I-Python environments using shared environments (Rosie, Google CoLab, etc.) and single-user environments (Anaconda)
• Python language review
  • Expressions, operators (arithmetic, logical and relational), data types (None, integers, floating point numeric and object), compound expressions leading to statements
  • Flow of control in Python (sequential, branching (if/elif/else), iteration (for loops, while loops, ranges, else parts to loops, iteration over collections), functions and lambda expressions). Points of syntax - importance of colons and tab stops
  • Collections in Python (lists, tuples, dictionaries, sets) indexing and slicing, iteration over collections using comprehensions, iterable objects and generators
  • Classes and object-oriented programming in Python (instance variables, methods, operator overloading, inheritance, polymorphism)
  • Numeric Python (NumPy), Scientific Python (SciPy) and Pandas Libraries
• Machine learning basics
  • Linear regression fundamentals, simple and multivariate linear models
  • Training sets, test sets, validation sets and other data set partitioning
  • Closed form equation solution for linear regression on small data sets
  • Iterative linear regression techniques for large data sets
  • Learning rates, error surfaces, residual analysis
  • Problems related to underfitting and overfitting
• Multivariate regression models
  • Polynomial regression and curve fitting techniques
  • Logistic regression for binary decision-making
• Decision model regularization
  • Occam’s razor and its role in model reduction
  • Ridge regression and Lasso regression
  • Model parameters and hyper-parameters
• Support Vector Machines (SVMs)
  • Linear and non-linear SVMs for Classification (SVCs)
  • SVM Regression (SVRs)
• Decision trees
  • Decision tree regressors
  • Decision tree classifiers
• Ensemble learning with random forests
  • Bagging and pasting
  • Boosting techniques
  • Gradient boosting
• Unsupervised learning
  • Clustering, k-nearest neighbors, k-means clustering
  • DBSCAN and other clustering algorithms
• Artificial neural networks
  • Artificial neurons and the digits problem
  • Perceptrons and multilayered neural networks
  • Backpropagation algorithm
• Deep learning neural networks
  • Using Keras as a front end to implement deep neural networks
    • TensorFlow - a deep learning library from Google
    • PyTorch - a deep learning library from Facebook
  • Image processing and computer vision models
  • Natural language processing models