The hardest interview question

What is your favorite class and why?

It should be an easy question. But for someone breaking into a new profession and being excited by all the options, it is the worst.

I’ve been answering the question by talking about one of the classes I am in now, as it is top of mind. I decided to reflect on my favorite parts of all the classes here to help me answer this question next time it comes up.

• Statistical Methods: Testing whether two groups are statistically different from one another opened up a whole world of questions to answer. When I had COVID, I did a little side project. It seemed like the men in the program talked for more minutes, on average, when giving summer presentations than the women. I recorded the times for each student using the recordings, subtracted the interruptions by faculty, and did a test. Men and women did not have a statistically significant difference in the length of speaking time. Hunches are not significant!
• Advanced R programming: We had to analyze messy form data about program participants. OK. So, I was responsible for half the chaotic inputs. When we were asked to fill in the form, I could soooo tell that we would analyze it. I wanted to practice dealing with complex to work with data!
  
  I wrote functions in R for the first time. This project was due over the July 4 holiday when I was at the beach. I was happily holed up indoors trying to make the output account for ties and put the months in monthly order rather than alphabetical for our output. I like making things tidy more than I like sitting on the beach!
• Advanced Python: 🐍I organized a “share-a-thon” after our first assignment was due. We each had to create a program to collect weather data from an API and calculate some statistics. We also had to output the results in a specific format. Once we all individually submitted our assignments, I invited people to share their codes over pizza. 🍕  I learned so many functions and options that I would not have guessed existed from this one-hour share-fest!
• Time Series Forecasting: ⏰ I am not a competitive person. Or I thought I wasn’t…. Seeing a time-varying pattern in data and being asked to predict the future…. oh yeah. That made me want to win! The assignment was to predict energy use into the future. We were introduced to multiple algorithms. I played with all the options on all of them in hopes of getting a bit more of an advantage! Ensembling and my ability to parse documentation were the winners. My free time on the weekends was the loser. Energy, real estate, or general demand forecasting, anyone?
•  Logistic Regression:  We had a multi-week project to predict which customers would purchase an annuity. The data exploration was one of the best parts of this assignment. I discovered a pattern, no doubt planted by our instructor, in the missing data that helped our team make an excellent business recommendation. Model building is fun, but time spent in thorough data exploration is well-spent. This class made me want to work for an investment company.
• Data Mining: ⛏ This class introduced so many techniques. I have been waiting for free time to apply them to new contexts. One of the ideas I loved best was the “nearest neighbors,” as I had been asked as a new graduate student in astrophysics to do some nearest neighbors analysis of galaxies. That project would have been much better/faster with some of the algorithmic oomph available in R today.
• Machine Learning: Many people like to toss things into a machine learning algorithm without checking assumptions the way you need to in linear regression. I liked the idea of tuning. I didn’t have enough time to tune everything optimally, but I liked the idea that I could find a better solution and that there were ways to tell you had a better solution. I don’t like guessing. I like knowing.
• Survival Analysis: We analyzed pump failures during Hurricane Katrina. This project made me want to work for an engineering company analyzing equipment or doing quality control analytics.
• Cloud Computing:   ☁ I used the cheap parallel computing method to finish my Ph.D. I borrowed other people’s Unix servers overnight and launched jobs, harvesting the results in the morning. As part of my cloud computing class exercise, I launched three servers for computing power that could have completed my Ph.D. simulations in minutes… I want more of this!
• Optimization: Setting up equations and letting a computer figure out the answer… What is not to love? The examples in this class reminded me of the logic puzzles I used to do in middle school. Maybe I shouldn’t post this publicly, but I might do this sort of work for free if I win the lottery!
• Financial Analytics: We got to build a scorecard to help make bank lending decisions! How many lenders have applied a scorecard to me for a mortgage or credit card? I finally got to peek into the secret sauce and am ready to help with credit decision-making. 💲
• Simulation and Risk: We projected whether an oil drilling project investment would be profitable. Getting the parts right and double-checking with another team member was a lot of work but worth it. 💰
• Customer Analytics: Why do web pages look like they do? And how can you measure the effects of a change to a web page? This class made me want to work ANYWHERE they do A/B testing. Disney? Bank? E-tailer?

Each time I am asked, “what was your favorite class” is, I have a different answer. Every class has been interesting. Every course has made me want to work in the industry related to the class examples.

My inner voice is asking me to get an answer together, so I don’t sound indecisive! Maybe I should try to focus my answer to this question on something I know won’t change as I learn new business cases?

I like validating. Do my answers agree with my teammates? I have been very conscientious about double-checking. Some may say, “It’s just homework.” But, concluding incorrect analysis is costly. I want to make sure the answer makes sense. This is a favorite activity that cuts across all the classes. And I’ve learned some people work as “model validators”…. 💯

That’s my new answer… if I can remember. We start some new classes next week

Pattern image by Vectorportal.com, CC