After dabbling in data science in her Analytics career, Deseanae wanted a career change without racking up debt for another degree (she already had a Bachelor’s in Engineering and an MBA). Enter Galvanize! Deseanae shares the differences between R and Python, her stellar final project and tips for other future data scientists on the interview circuit.
What were you doing before you came to Galvanize?
I have a background in engineering and an MBA. Most recently I was doing business and market analytics. I enjoyed using data-driven models to discover insights.
I started looking at related positions and found that most positions required a PhD in statistics or coding experience. But, I was already building models in R and I’m the type of person that can teach myself anything. However I quickly realized there was a lot more to learn than could be feasibly self taught. So I looked at my options and was actually considering a second masters.
Thinking about going back to school, what was that decision making process like?
I was thinking about a masters in statistics or data science, which is a branch of statistics. It was a hard decision because I have a monthly student loan payment that is bigger than most people’s mortgage. It was really hard for me to consider taking on more debt.
One of my coworkers told me about Galvanize. I was already thinking about moving to Denver and when I moved here a neighbor mentioned it too. So I checked it out, and the data science program seemed like the additional step that I needed.
It is a hard decision. The Galvanize program in Denver is $16,000. It is expensive but it’s a lot less than 140K. One thing I will say, you do learn a lot very quickly and we are here learning 11 hours a day. Even though it’s 12 weeks, if you compare that to a college program, it’s probably three semesters of school.
What was the application process like?
There’s a written application, two coding assignments and a statistics assignment. There’s a coding assignment that you take home and another coding and statistics assignment during a Skype or Google Hangout where they’re actually Skyping with you and watching you code – which is a little unnerving.
You mentioned R. Is that the language you learned in the Galvanize class?
No, we learned Python, which I crammed in about two hours. I learned enough to take the test. A lot of my classmates had a lot more than two hours of Python. The first week was very hard for me just trying to pick up the language in addition to picking up the statistics.
There are a couple of data science schools in New York that teach R. What is the difference between R and Python?
They’re both open source, but R’s a lot older. The thing with R is because it is open source and older there are a lot of libraries but it’s not really standardized. You use one library to do something one day and a week later, you come back and you can’t.
Python, which is newer, is more robust. Even though it only has a couple of libraries, the libraries are very well-defined, very well documented. When there’s an issue, you can upload your issue and the people who are writing it will fix it and do the update. You don’t have to change your code, which is why people are moving to Python, because it’s just stronger.
Can you do things in Python that you can’t do in R?
You can do things easier in Python because it has stronger libraries. The big library that people use is Panda’s. In Panda’s library, I can do one line that would have taken several lines of code in R.
Another big library is SK-Learn which is like a statistical package. Someone with a math background might take three pages front and back to do a task, but the same task would only require a few lines of code in SK-Learn. The library is so robust that one line can accomplish a lot of things!
What other technologies and frameworks did you learn at Galvanize Data Science?
We used Hadoop, SPARK, Amazon Cloud, EC2 and more. Everything was based in Python. We built onto Python then we moved on to Hadoop and SPARK for the larger data sets.
Since you have many years of traditional education, what was the difference between your undergrad program and your 12 weeks at Galvanize in terms of teaching style?
I like the teaching style at Galvanize better. It might be because I already have the foundation; I’m not a bright-eyed 18-year old anymore. I went to Cornell for undergrad and I remember my entry level engineering classes felt overwhelming. I was just completely lost because the classes are so big and you don’t feel comfortable asking questions, that’s just part of being young and naïve.
At Galvanize, since our class was small and our professor was in the industry, if I didn’t understand something I could ask the professor to explain it in more detail or I would ask my classmates, “How did you learn this?” They would either bring in a textbook the next day for me to read those chapters or they would email me an article that explained it from a different approach. In that way, I really thought it was different.
Were there women in your Cornell undergrad engineering program?
Yes, there were always one or two girls.
Was the class at Galvanize a diverse class?
No, I’m the only girl.
Who was the instructor for this class and do you know their background?
Dan Becker; his background is in economics and statistics. He started doing Kaggle competitions. In a Kaggle competition, a company poses a problem for a group to work on and the group or individual that solves the problem can make really good money, about 50 to 100 K.
So, he started winning Kaggle competitions and getting job opportunities, which led him to data science. From that he started doing consulting work and now he’s teaching. He's transitioning to a startup role so we’ll be getting a new teacher.
What did you do for your final project?
I created a predictive model that utilized gender, education level and other demographic factors to determine the probability of a customer buying a particular product with 98% accuracy. I was able to do this using machine learning.
If you’re using your gut to predict consumer behavior, which salespeople use a lot, there’s no way to quantify that. Whenever I used Excel and Excel Miner the highest accuracy rate I achieved with real data was 70%.
After I created the model, I created an interactive dashboard hosted in an app. If my sales people were meeting customers in the field, they could easily type in their demographic information and determine if a customer is going to buy a product and what product features are important to them.
What was the biggest challenge of these 12 weeks?
I think the biggest challenge, which might sound very trivial, is learning how to use an Apple computer.
What are you looking at after this?
I’m looking for jobs right now, which is the scary part. When you graduate from a university, they have everything kind of lined up for you.
Here we have connections with companies, but they don’t have positions set aside or developed. I want to find a position that combines my management skills with business and data science. I would like to find a big data product management position or a data science marketing/project management position. Right now I’m sending applications to get in front of people and explain my skill set, because I feel like a lot of times people are looking for certain keywords. Because I have such a diverse background talking to them allows me to explain how everything lines up and how I have all of the necessary skills.
Have you gone on an interview yet?
I’m interviewing with one company now, but they’ve put their interviews on hold. It’s a consulting company that works with data science solutions for external companies.
Do you have any advice for people who are looking at data science boot camps in general or people who are looking at Galvanize specifically?
Though we’re using data science as a statistical base, everything still has to do with code. Definitely brush up on your coding skills or buy a Python or Linux book. I bought Intro to Python.
I actually bought a SQL book also and am learning to answer CS questions because there’s a misconception about what data scientists are really doing. A lot of interviewers ask CS or SQL questions even though in your role you would never use SQL or code as a programmer does. You’re not going to be using SQL unless you’re a database manager which is a completely different skill set.
Any general advice to bootcampers?
The last thing I would say is because these bootcamps are so fast and it’s like going into a new world, especially if you’re not in New York where it’s crazy exploding, I would say take the time to go to meetups and talk to people.
Before you know it you’re going to be done with the program and you haven’t made those connections to see who’s actually hiring and what they’re looking for. Galvanize has a lot of meetups in their space.