Rolling Forward! Data Science Boot Camp Wk 2–4 Update.

David Lee
7 min readAug 20, 2020

“This is the most difficult program General Assembly offers,” says my cohort’s dedicated career outcomes coach.

The past three weeks have been tough for me, as I had quizzes, project deadlines, and no power and internet for four days as Tropical Storm Isaias swept through Connecticut leaving 600,000 people in the dark. While most wouldn’t dare showcasing personal failure in a public space, my story has come out of tumbling forward through each trial, and sharing highs and lows along the way towards becoming a Data Scientist.

When falling forward isn’t fast enough, Wk2:

Quiz 1 Statistics: Mean: 12.66pts –25th Percentile- 9pts –50th Percentile: 12pts –75th Percentile: 18pts. The passing score on this quiz is 12 points. Sourced from our Google Classroom notes.

Lets be real:

I remember speaking with my wife right before my class at GA about the intensity of the program based on my research along with the possibilities of falling too far behind in academia to pass. If you have read through my journey from previous posts, you’ll have a good idea to my limited exposure to the range of topics covered in Data Science.

Having said that, there are times in my life where I could have focused on, tried harder, and spent more time on developing my craft. Knowing the challenges I face in this boot camp, I never want to put myself in a position again where I question myself if I could have achieved my goal if I had given it my all.

My inexperience in Math and coding really showed on week 2 as I received a failing score on my first quiz, and my progress speed on lab assignments were about half of most of my peers. I averaged nearly 4 hours of sleep per day for fear of missing my large project deadline on Friday, July 31st. Come Thursday night close to midnight, I had finally just cleaned my data set for Project 1, and still needed to complete several visual analysis, create and answer a problem statement through insights I made on the data, along with preparing a 10 minute slide presentation that I had to virtually present with 10 hours to go. I confidently gave this every ounce of my strength, and I came to the realization that my best still might not be enough.

As I’m trying to think of a way to miraculously pass my first major project, I knew that even if I spent the next 10 hours starting my visual analysis on my data, my coding skills were not good enough to finish on time. Passing every Biweekly project is a requirement to pass the course with a GA certification of completion so I was very concerned and tired from lack of rest.

Get up!

A helping hand, in my time of need! Thank you..

One of my friends in my cohort helped me find my only way out, and it didn’t involve giving in to the comfort of my bed.

The past catches up

Although the program emphasizes the use of the programming tools we are learning, I know very well the importance of meeting a deadline. With experience in RFPs (Requests for Proposals) for Board of Education systems and State and Local Government, I know that if you are even a minute late for a bid, your weeks or months worth of proposal prep is thrown out. Simply put, be prepared to be on time with good information!

Now past the point of completing the project conventionally, I had to resort to gaining insights on my data the most efficient way I could with hours to go. I did my analysis on Microsoft Excel, make visuals with basic statistics, gaining insights on my data. I further researched my topic on the web, to learn more about potential problems my findings can solve, and I created my presentation on Google Slides.

I presented my project after napping for 2 hours with only the information shown on my slides as I didn’t have time to prepare notes. After class, I went back to my project notes, and completed the majority of my visual analysis before the deadline that day. I fully expected to fail my project, which I would have been OK with as I couldn’t have given it more effort than I have.

Fast Forward to weeks 3 and 4

I had to inform my Instructors of my present situation.

Week 3 consisted of another cycle of events. I finally overcame my obstacles on Project 1, and was excited to be closer to my classmates on labs and the lessons until Tropical Storm Isaias came on Tuesday August 4th. I didn’t really know what to do and was anticipating an update from my electric service provider. Losing a single day in a coding boot camp is a BIG deal, and three absences in my program results in failing the class! To my dismay, my service provider sent me an update the next morning saying that the restoration would likely happen close to the weekend!

Another test

Brainstorming with friends!

Thinking about my situation, I’ve considered my options. I seriously considered having to drop my class and go the self study route as I knew the pace of information I would be missing in class, and my WiFi options were few. Thankfully, the recommendation of hotels in the above text gave me my only possible solution, which was my local church.

On Wednesday, August 5th, I received permission from the church to study using their WiFi, and did my classes and labs until power was restored on Saturday, August 8th.

Roll forward if you are going to fall!

My experience in GA’s bootcamp has been tough so far! So many things that could have gone wrong has. I remember back when I was a kid learning Taekwondo, we would learn rolls and breaking falls. The point was to prevent enemy attacks, get up quickly, and prepare for a counter. To fall forward, and take the time to get back up might be too slow in this type of learning setting, and in my case, you need to be prepared for anything. The sad truth is that not everyone can learn at this pace, which I have already seen, but I’m continuing to do my best to show that someone like me still has a chance.

Some highlights:

I passed my Quiz 1 retake, a new set of problems.
I passed Project 1.

I think Project 2 also went well, and it was my first time using Machine Learning with a first Multi-Linear Regression Model predicting home prices on unseen data.

Training a prediction model with nearly a 90% accuracy on my train/test data, and about a 85% accuracy on new data on a private classroom Kaggle competition. Model made with Scikit-learn and Yellowbrick.

Wrapping Up

Every moment up until this point has felt like holding on to a rope tied to a jet ski, falling while never letting go. I received a lot of great help and advice from my peers and instructors who have pulled me forward. Most importantly, my family has accommodated me so I can focus on maximizing my educational experience at General Assembly. Thank you deeply to everyone involved!

A Team Filling the Gaps

I could not have gone this far on my own. There are areas that I easily identify as slow and steady developments for me. I know I’m improving with the several hours of daily practice, but all of my friends here have saved me from trouble. Each of us have different abilities that shine, and its been amazing how we multiply our learning as we study together and solve difficult problems. This I expect is exactly what a good Data Science team looks like.

There is a term in this field that we look for in our models called Multi collinearity. This is when variables that are too highly correlated with each other are included in predicting our result, resulting in a worse model. The fix to this problem is to remove multiple instances of the variable or to combine them so the weight compared to other variables is fair, creating a better performing algorithm. Its very similar to the amazing different skill sets that our team has established, and we are constantly sharpening each others skills making sure we learn as efficiently as possible.

Next Steps

For the first time since the start of the program I finally feel like I’m one step ahead of the program. I’ve completed my lab early this week, made progress on my next major project, and I’m able to contribute more to my team. I think I’m ready to tackle my Data Science project with the intent on making it portfolio ready, a machine learning one I can finally be proud of.