1. Home
2. Faculty of Arts and Sciences

STAT571 DATA MINING AND MACHINE LEARNING

Course Objectives

Understanding Data Mining Fundamentals: Gain a comprehensive understanding of the principles and techniques of data mining. Explore the theoretical foundations underlying data mining methodologies.

Practical Application of Data Mining Methods: Develop practical skills in applying various data mining methods to real-life problems. Understand how to preprocess data effectively for mining tasks.

Evaluation and Critique of Data Mining Approaches: Critically evaluate the strengths and limitations of different data mining techniques. Assess the suitability of specific methods for solving different types of problems.

Mastery of Machine Learning Concepts: Master essential concepts in machine learning, including supervised and unsupervised learning paradigms. Understand the principles behind popular machine learning algorithms.

Hands-on Experience with Model Validation Techniques: Gain practical experience in validating and evaluating machine learning models. Learn techniques for assessing model performance and generalization capabilities.

Exploration of Advanced Topics: Explore advanced topics in machine learning, such as ensemble methods, deep learning, and generative adversarial networks. Understand the theoretical underpinnings of these advanced techniques.

Practical Implementation Skills: Develop practical implementation skills through hands-on programming assignments. Gain proficiency in using popular libraries and frameworks for data mining and machine learning.

Application to Real-World Scenarios: Apply data mining and machine learning techniques to real-world datasets and problems. Understand the challenges and considerations involved in applying these techniques to practical problems.

Communication and Presentation Skills: Develop effective communication skills for presenting and explaining data mining and machine learning concepts. Learn how to interpret and communicate the results of data analysis effectively.

Ethical and Responsible Data Mining Practices: Understand the ethical implications of data mining and machine learning. Learn about best practices for ensuring fairness, transparency, and accountability in data-driven decision-making.

By the end of the course, students will have the knowledge, skills, and practical experience necessary to apply data mining and machine learning techniques effectively to solve real-world problems and contribute to advancements in the field.

Course Content

Unsupervised learning. Principal component analysis (PCA), clustering methods. Rule learning, association rules. Supervised learning. Multiple linear regression. K-nearest neighbors. Logistic regression. Linear discriminant analysis. Linear model selection. Regularization techniques. Ridge regression, LASSO. Splines. Generalized additive models (GAMs). Tree-based methods. Ensemble learning. Bagging, random forest, boosting. Support vector machines. Neural networks and deep learning. Evaluating the performance of machine learning algorithms. No Free Lunch theorems. Bias-variance decomposition. Bagging. Boosting. Generative adversarial networks (GANs). Autoencoders. Variational Autoencoders.

Course Learning Outcomes

Evaluate Data Mining Applications: Critically evaluate the value and application of data mining techniques for addressing real-life problems across various domains.

Comprehensive Understanding of Data Mining Methods: Explore diverse methods in data mining, encompassing data analysis, statistical techniques, machine learning algorithms, and model validation procedures.

Understand and apply fundamental modeling approaches including linear regression, linear classifiers, decision tree models, and clustering algorithms.

Explain the No-Free-Lunch theorems and elucidate the significance of prior knowledge in solving machine learning problems effectively.

Bias-Variance Analysis and Regularization: Derive the bias-variance decomposition for Mean Squared Error (MSE) and "0-1" losses, and illustrate how regularization impacts the bias-variance tradeoff.

Ensemble Learning Techniques: Explain bootstrapping, bagging, and boosting concepts, and justify the selection of weak learners for specific aggregating algorithms.

Relationship Between Linear Models and Deep Neural Networks: Discuss the connection between linear models and deep neural networks, and describe the training process of neural networks.

Understand the principles of Generative Adversarial Networks, identify metrics they optimize, and explore techniques for regularization.

Handling Imbalanced Datasets: Apply techniques for effectively working with imbalanced datasets, ensuring robustness in model training and evaluation.

Knowledge Discovery and Data Mining Tasks: Perform essential computational tasks of data mining such as pattern extraction, association mining, classification, clustering, ranking, prediction, and outlier detection.

Demonstrate the formulation and representation of real-world applications as different types of data, including matrices, itemsets, sequences, time series, data streams, etc.

Identify appropriate data mining techniques for specific real-world scenarios, considering the nature of the data and the problem at hand.

Apply state-of-the-art data mining techniques to address various problems across different domains effectively.

Develop software development skills necessary for handling large-scale datasets, including data preprocessing, model development, and evaluation, considering datasets with millions of records.

Program Outcomes Matrix

0: No Contribution 1: Little Contribution 2: Partial Contribution 3: Full Contribution