Description of Course Goals and Curriculum
Main points:
- Often called a “machine-learning course for humanities majors”
- Focuses on all stages in the data analytics process
- Curriculum involves both theory and coding of machine learning methods
The course is a relatively comprehensive overview about the entire process of predictive data analytics, including identifying the problem, examining the data, crafting a high-level solution, different types of machine learning methods that may be useful, and evaluating the machine learning model. The curriculum covers a wide variety of different types of learning methods, including the theory behind the models using linear algebra and calculus, and the very basics of coding these learning methods using Python. Although coding and math (201 and 202) experience can be helpful, it is not vital to learning the material, and the goal of the course is to understand the entirety of using machine learning to help with data analytics in a business setting.
Learning From Classroom Instruction
The course was taught entirely in a large lecture-based format, twice a week. Dr. Jha is a great lecturer who goes slowly and teaches the material comprehensively. Each lesson is based off of chapters in the textbook for the course, and the slides also correspond with the chapters, so it is extremely easy to match up what you learn in lecture and what you’ve read in preparation for the class. The most important thing is reading the textbook, as the lectures are based straight off of that information, and usually only cover the most important points; however, future assessments and assignments may require more extensive knowledge that can be found only in the textbook. Dr. Jha provides the slides for the textbook in advance, and it may be helpful to either review the slides beforehand or write directly on the PowerPoint slides when taking notes during lecture. Dr. Jha also loves to incorporate student participation and will often ask questions for the class as a whole to answer, so if you have a question, don’t hesitate to ask! The professor is extremely open to hearing the students’ feedback, and hearing about which parts of the lecture are particularly difficult to understand.Learning For and From Assignments
There are four main different types of assignments and assessments that I would like to discuss: quizzes, homework assignments, coding projects, and exams.
In order to make sure we are keeping up with the material and understanding lectures, we are given quizzes in class on a biweekly basis. Although these quizzes are technically impromptu and unannounced, they are usually given around the same time that the homework assignments are scheduled to be submitted, and cover material that was on that week’s homework. The quizzes are also worth relatively little in terms of actual grading, so the most important thing is to check your own understanding of the material and nail down what you are confused about. There is no point in stressing unnecessarily. Many times, these quizzes will cover problems similar to the homework, simple terms or concepts from the textbook, or details from lecture. Therefore, the best way to study for these quizzes is to do the homeworks that are assigned, do the readings and listen in class. Immediately after, Dr. Jha will go over the answers for the quiz, so you can feel free to ask questions after class if there are particular points of confusion.
Then, we have homework assignments that are due approximately every other week. They will cover different topics from lectures and the textbooks. Most of the time, even if you are confused at the beginning, the homework assignments are very doable if you use the textbook as a reference, paying particular attention to the examples. We may also have some more mathematical exercises to do, which may require knowledge of linear algebra and calculus. These homework assignments focus on the theoretical understanding of different types of learning methods, and also the general goals for data analytics. We are allowed to work in groups, and collaboration can help you digest the material better. Furthermore, office hours are a great resource if there is anything that is confusing.
In the middle of the course, we begin learning about different types of machine learning methods, during which we get programming projects to complete. Each of the four types of learning methods have an associated coding project, where we fill out a template of shell code to run a simple analysis on a data set. This is a cool part of the class because it helps cement our understanding of the theoretical underpinnings of the method, but also allows us to see the practical applications of what we have learned. We get a certain data set, for example, a dataset about diabetes, and we can run analysis to predict certain aspects and characteristics for a subject using what we have learned. Although this may sound daunting if you have no prior coding knowledge, I went in without any Python experience and didn’t have too much trouble. The projects themselves typically assume that you have little or no experience, and a lot of it is already written, leaving you to fill in certain parts.
Additionally, office hours are a great resource if you’re stuck on your code.Lastly, we have the actual exams, which is one midterm and one final. The exams can be considered a big amalgam of the quizzes and the homeworks, and usually time is not a large issue on these exams. The exams tend to have both smaller-point questions which focus on simpler concepts and terms, and larger-point questions that are more involved, which may require more math or understanding of an entire learning method from start to finish. While the exams may be daunting, I think that completing the entire course (doing readings, taking the quizzes, doing homeworks) is good preparation for the exam, and there is not much material outside of what you have already seen in classes and on prior assignments. Furthermore, we are usually given a practice exam, although without the answer key, that we can use to practice prior to the final. Furthermore, many of the homework problems actually come from prior examinations, which shows how helpful doing these assignments can be for studying for the final. The best way to prepare is reviewing old problem sets and doing the practice!