On April 1st, we hosted an AMA (Ask Me Anything) session on our Community Slack channel with Sergey Fogelson, Vice President of Analytics and Measurement Sciences at Viacom and instructor of our upcoming Introduction to Data Science course. He developed this course and has been teaching it at Metis since 2015.
During the AMA, attendees asked Fogelson questions about the course including what to expect, how to prepare, and how it will benefit their goals and careers. Read below for some highlights from the hour-long chat.
This Introduction to Data Science course starts Monday, April 8th and runs through May 16th on Mondays and Thursdays from 6:30 - 9:30 pm CST. Learn more and enroll here.
What can we reasonably expect to take away by the end of this course?
The ability to build a supervised machine learning model end-to-end. So, you'll be able to take some data, pre-process it, and then create a model to predict something useful by using that model. You'll also be armed with the basic skills necessary to enter a data science competition like any of the Kaggle competitions.
How much Python experience is necessary to take the Intro to Data Science course?
I recommend that students who want to take this course have a bit of Python experience before the course starts. This means spending a couple of hours of Python on Codeacademy or another free resource that provides some Python basics. If you're a complete novice and have never seen Python before the first day of class, you're going to be a bit overwhelmed, so even just dipping your toe into the Python waters will ease your path to learning during the course significantly.
I am curious about the basic statistical & mathematical foundations part of the course curriculum – can you expand a little on that?
In this course, we cover (very briefly) the basics of linear algebra and statistics. This means about 3 hours to cover vectors, matrices, matrix/vector operations, and mean/median/mode/standard deviation/correlation/covariance and some common statistical distributions. Other than that, we're focused on machine learning and Python.
Is this course better seen as a standalone course or a prep course for the immersive bootcamp?
There are currently two bootcamp prep courses offered at Metis. (I teach both courses). Intro to Data Science gives you an overview of the topics covered in the bootcamp but not at the same level of detail. It is effectively a way for you to "test drive" the bootcamp, or to take an introductory data science/machine learning course that covers the basics of what data scientists do. So, to answer your question, it can be treated as a standalone course for someone who wants to understand what data science is and how it's done, but it's also an effective introduction to the topics covered in the bootcamp. [Here is a handy way to compare all course options at Metis.]
As an instructor of both the Beginner Python & Math course and the Intro to Data Science course, do you think students benefit from taking both? Are there major differences?
Yes, students can definitely benefit from taking both and each is a very different course. There is a bit of overlap, but for the most part, the courses are very different. Beginner Python & Math is about Python and theoretical basics of linear algebra, calculus, and statistics and probability, but using Python to understand them. It's really the course to take to get prepared for a bootcamp entrance interview. The Intro to Data Science course is mainly practical data science instruction, covering how different models work, how different techniques work, etc. and is much more in line with day-to-day data science work (or at least the kind of day-to-day data science I do).
What is suggested in terms of an outside-of-class time commitment for this course?
The only time we have any homework is during week 2 when we dive into using Pandas, a tabular data manipulation library. The goal of that homework is to get you familiar with the way Pandas works so that it becomes easy for you to understand how it can be used. I would say if you commit to doing the homework, I would expect that it would take you ~5 hrs. Otherwise, there is no outside-of-class time commitment, other than reviewing the lecture materials.
If a student has extra time during the course, do you have any suggested work they can do?
I would recommend that they keep practicing Python, like doing additional exercises in Learn Python the Hard Way or some extra practice on Codeacademy. Or implement one of the exercises in Automate the Boring Stuff with Python. In terms of data science, I recommend working through this grandaddy-of-them-all book to really understand the foundational, theoretical concepts.
Will video recordings of all the lectures be available for students who miss a course?
Yes, all lectures are recorded using Zoom, and students can either rewatch them within the Zoom interface for 30 days following the lecture or download the videos via Zoom directly to their computers for offline viewing.
Is there a viable path from data science (specifically starting with this course + the data science bootcamp) to a Ph.D. in computational neuroscience? Said another way, do the concepts taught in both this course and the bootcamp help prepare for an application to a Ph.D. program?
That's a great and very interesting question and is much the opposite of what most people would think about doing. (I went from a Ph.D. in computational neuroscience to industry). Also, yes, many of the concepts taught in the bootcamp and in this course would serve you well in computational neuroscience, especially if you use machine learning techniques to inform the computational study of neural circuits, etc. A former student of one of my Intro course wound up enrolling in a Psychology Ph.D. after the course, so it's definitely a viable path.
Is it possible to be a really good data scientist without a Ph.D.?
Yes, of course! In general, a Ph.D. is meant for someone to advance some basic aspect of a given discipline, not to "make it" as a data scientist. A good data scientist is simply a person who is a competent coder, statistician, and fundamental curiosity. You really don't need an advanced degree. What you need is grit, and a desire to learn and get your hands dirty with data. If you have that, you will become an enviably competent data scientist.
What are you most proud of as a data scientist? Have you worked on any projects that saved your company significant money?
At the last company I worked for, we saved the firm a significant amount of money, but I'm not particularly proud of it because we just automated a task that used to be done by people. In terms of what I am most proud of, it's a project I recently worked on, where I was able to forecast expected ratings across our channels at Viacom with much greater precision than we had been able to do in the past. Being able to do that well has given Viacom the ability to understand what their expected revenues will be in the future, which allows them to make better long-term decisions.
To enroll in the Intro to Data Science course with Sergey, visit here.