Over the past decade, machine learning has emerged as an important research area with practical applications in a range of fields, including engineering and commerce. Many companies and researchers are using the abundant data available today to recommend purchases to consumers, predict the results of business actions and inform better decision making. Examples can be found in industries as diverse as retail, robotics, manufacturing and social networking. However, those using machine learning often face challenges implementing the right techniques for a particular situation, due to the dynamic nature of the field, in which new techniques are introduced frequently. In this course, we emphasize the fundamentals and prepare you to select and make use of appropriate techniques.
The course employs hands-on activities and practical examples to understand, build and test machine learning techniques. There will be a broad introduction to the key learning methods, including regression, classification, clustering and recommender systems. The learning theory is covered with an emphasis on its practical uses. The focus is not on individual algorithms but rather the ideas that make them work. The course reviews the steps involved in building predictive models, including data collection, feature selection, algorithms and evaluation. You will learn how to fine tune the performance of these models, and plan for practical implementation issues. Important topics will be demonstrated using real-world applications and case studies.
By the end of the course, you will have a basic understanding of machine learning techniques and know how to apply basic machine learning tools in practical situations. Students will complete several assignments and a project using supervised or unsupervised learning algorithms.
- Basic concepts in machine learning
- Review of linear algebra, R and machine learning tools
- Managing and understanding data, data import and export, and the structure of data
- Understanding classification using Nearest Neighbors, Naïve Bayes, and Decision Trees
- Forecasting: regression methods, black box methods, neural networks and support vector methods
- Finding patterns: market basket analysis using association rules
- Finding groups of data: clustering
- Time series analysis and mining
- Evaluating and improving model performance
- “Overfitting” and regularization
- Applying machine learning: guidance and practical issues
- Dimensionality reduction
- Detecting anomalies and outliers
Skills Needed: Some programming experience is recommended. R will be used in class examples, and Python experience can be helpful. Basic knowledge of probability and statistics is required. Prior machine learning knowledge is recommended but not required.