DS 3001: Foundations of Machine Learning
Your Foundations of Machine Learning Tour Guides:
Brian’s Office Hours: In person in my office Elson (165) or Virtual on Discord: Friday, 2:00 to 4:00 or by appointment if needed. Feel free to just come by and hang out if you’d like.
Ariful office hours on Thursday from 1:00 to 3:00 in Elliewood (above Ragged Mountain Running Shop on the Corner) small conference 3a
Course Materials: Foundations of Machine Learning Repo
Subject Area and Catalog Number: Data Science, DS 3001
Year, Term and Time: 2022, Fall, Thursday 3:30-6:00
Class Title: Foundations of Machine Learning
Credit Type: Grade (A-F)
What is Data Science, why is it becoming so important and what is needed to be successful in this field? We will explore these questions throughout the course through a variety of topics with data always at the center. In all things, we focus on creative thinking, not blind implementation. If you cannot answer why you are doing something, not only will you discover no new knowledge but you will also create new problems versus solving them.
The course centers on lab-based work and employs a team-based pedagogy, meaning much of the work in the course can and should be completed in collaboration with your classmates. Though very applied, we also include theoretical content and will have discussion sessions depending on the topic for any given week.
I am also aware that students come from different backgrounds and have a variety of skill sets that they bring to the class. There will be plenty of opportunities for those that would like to take advantage of extra discussions during office hours, TA support sessions, and team-based work that is designed in such a way that we can all learn from each other.
Throughout the course, we will endeavor to “live the life of a data scientist” allowing you to not only be directly taught but gain a sense of what it would be like to be working as a data scientist. You’ll be asked to discover new knowledge and share it with your peers and learn how to use the larger ecosystem of data science to your benefit.
Data Science is incredibly broad and dynamic. The topics below are designed to reinforce this perspective and help you understand the field’s core tenements and what is demanded from practicing data scientists. The key is for you to gain a sense of the scope of Data Science, what is needed to contribute to the community, and feel comfortable incorporating these techniques into your work moving forward. Specific learning objectives are below:
Be able to describe the field of Data Science and its emerging sub-fields Gain experience working in teams to solve Data Science problems Gain experience communicating Data Science products Articulate the advantages and disadvantages of selected ML approaches Be able to select appropriate ML models given problems and data types Understand the importance of and methods for evaluating ML models Understand the negative outcomes associated with ML/AI bias and how they can be avoided
The course will move rather quickly and can be demanding at times. However, if we all work together to support each other you’ll be amazed how much you learn at the end of the semester!
On any given week, the course will require reviewing short video lectures and completing readings prior to coming to class. These lectures and readings will then be implemented in the lab portion of the course which will be conducted during the scheduled class period. Lab sessions will include a variety of activities but mostly be centered on team-oriented coding assignments. Students can also use lab sessions to work on mid-term and final projects when needed.
Ethical data scientist reflection (0%, 10% to the lab, 5% to the final) – Based on in-class discussion and readings students will write a personal reflection of how ethics and data science interact and how we all can think about these concerns moving forward. This is critical to the work of a data scientist as the models we build have the potential to impact the lives of thousands of people so we must use caution and constraint whenever possible. Quizzes (0% 5% to the lab, 10% to the final) – Short occasional (5 or 6) quizzes, will be auto-graded, so you will get instant feedback. In order to ensure we are all meeting the learning objectives from week to week short quizzes will be given. You will be allowed as many chances to complete the quiz as needed and they will be open note however students are to work independently. Labs (60%) – On most weeks we will have in-class labs/assignments. These are designed to allow you to practice the skills being presented in class. While they should be submitted individually you are encouraged to work with your peers as much of the best learning can come from your peers. You’ll need to create publishable markdown documents for every lab and submit them along with the raw code file and link each week (Groups). Final project (40%) – The course will culminate in a final project that will involve working with a dataset of your choice, giving a presentation, submitting well-annotated code to include summary information in report form. This is an open-ended project designed to allow groups to choose a topic of interest from the semester to explore deeper and share with the class.
R/Rstudio - You'll need to have the software loaded and ready to go day one. Zoom - Virtual Option for Office Hours Github - Almost all course materials and collaberation for lab assignemtns Collab - Submission of assignments and class-wide communications Discord - Low latency comms for groups and class.
The books below are essentially a starter Machine Learning Library. I will use all of these references at difference points during the class, but not heavily rely on anything but the free options. Applied Predictive Learning is a theory and code review for much of what is in the CARET package. It’s a very good book but also expensive. Everything else is either free or can be found for around 15 dollars. If you can master what is in books F through I, you’ll be well on your way to being a highly capable Data Scientist. In order of importance I would list them in reverse, I-F.
***NOTE: depending on student interest, the syllabus can be adjusted to accommodate additional topics
|Week||Theme||Topics||Lab||Reading/Repo (Prior to Class)|
|Week 1||What is this “Data Science” that you speak of and tech stack||- Assessment - Videos: DS Overview and History||- Find DS Dream Job - Create your first project, load the dataset, visualize using the code provided what questions could this data answer?||Synchronous: Short Lab|
|Week 2||Getting back up to “coding speed”||‘Dataframing’ with tidyverse functions||- Group Case Study - Questions + PsuedoCode + Code + Functions = High Quality Data Science||“Program” Chapter in C|
|Week 3||How to share nicely||Knitr and rmarkdown||Rmarkdown article reviews||Rmarkdown: The definitive guide: Sections I and II Knit R Reference Guide|
|Week 4||Introduction to ML Concepts I||Language of ML||Case Studies||H: Chapter 1 and 2|
|Week 5||Introduction to ML Concepts II||Data Preparation:kNN||ML Concepts||H: 3 and 4|
|Week 6||Introduction to ML Concepts III||Machine Learning Process:kNN||ML Concepts||H: Chapters 3 and 4|
|Week 7||Spring Break|
|Week 8||Introduction to ML Concepts IV||Evaluation||Evaluation Lab||All of B. and G.- Chapter 11|
|Week 9||Nature’s Perfect ML analogy: Trees Part I||Classification: Decisions Trees||Decision Trees||F. Chapter 5 and G. Chapter 14.1-14.3|
|Week 10||Nature’s Perfect ML analogy: Trees Part II||Regression: Decision Trees||[Predicting Income for Big Brother]||F. Chapter 5 and G. Chapter 8|
|Week 11||Wisdom of the Crowd||Ensemble Methods I||Random Forest Classifier||TBD|
|Week 12||Kaggle Competition|
|Week 13||Let’s gather together… but separately||Unsupervised: Overview of Clustering Kmeans||NBA Scout for the worst team in the league||F. Chapter 1 and Chapter 9|
|Week 14||Do the next right thing…ethics||Bias in AI Discussion -Simple methods for identifying bias - Protected Classes||Fairness Overview & Ethical Reflections||Weapons of Math Destruction|
|Week 15||Final Project Prep||Final Project Overview||Ethical Reflection Due|
|Week 16 - Final TBD||Final Projects Presentations||Final Project Overview|
Grading Policies: Courses carrying a Data Science subject area use the following grading system: A, A-; B+, B, B-; C+, C, C-; D+, D, D-; F. The symbol W is used when a student officially drops a course before its completion or if the student withdraws from an academic program of the University.
University of Virginia Honor System: All work should be pledged in the spirit of the Honor System at the University of Virginia. The instructor will indicate which assignments and activities are to be done individually and which permit collaboration. The following pledge should be written out at the end of all quizzes, examinations, individual assignments, and papers: “I pledge that I have neither given nor received help on this examination (quiz, assignment, etc.)”. The pledge must be signed by the student. For more information, visit www.virginia.edu/honor.
Special Needs: The University of Virginia accommodates students with disabilities. Any SCPS student with a disability who needs accommodation (e.g., in arrangements for seating, extended time for examinations, or note-taking, etc.), should contact the Student Disability Access Center (SDAC) and provide them with appropriate medical or psychological documentation of his/her condition. Once accommodations are approved, just follow up with me concerning any logistics and implementation of accommodations. Please try to make accommodations for test-taking at least 14 business days in advance of the date of the test(s). Students with disabilities are encouraged to contact the SDAC: 434-243-5180/Voice, 434-465-6579/Video Phone, 434-243-5188/Fax. Further policies and statements are available at www.virginia.edu/studenthealth/sdac/sdac.html
Technical Support Contacts
Login/Password: email@example.com UVaCollab: firstname.lastname@example.org BbCollaborate Support: http://www.tinyurl.com/uvabbc