Inside the U.S. Natural Disaster Predictor Built by 4Geeks Academy Grads

Recent Data Science Bootcamp graduates Kelsey Low and Trevor Kortum put their skills to the test by building the U.S. Natural Disaster Predictions project at 4Geeks Academy. Their model analyzes historical data to predict the likelihood of natural disasters across the country—a project Trevor worked on while sheltering in place during the Los Angeles wildfires. In this spotlight, they break down how they built their predictor, the challenges they faced, and the key takeaways from their bootcamp experience. Learn how 4Geeks Academy prepared them for in-demand data science roles and their top advice for incoming students looking to make the most of the program.

Trevor, what inspired you to get into data science?

I come from a film background and for a long time, I didn't realize how much I was drawn to the analytical side of the industry. I was always fascinated by box office numbers and how some movies would exceed expectations while others underperformed despite big budgets or A-list talent. I found myself constantly looking at trends and comparing production costs, casting choices, and marketing strategies to see how they impacted a film's overall success. At the time, I didn't think of it as data science; I just thought I was deeply invested in the business side of film. Looking back, I was trying to build mental models for box office projections without realizing it. That curiosity naturally translated into my interest in data science.

Kelsey, why did you decide to go from software engineering to data science?

Over the last few years, I've seen the need for software engineers grow stagnant and decrease, whereas data science and particularly machine learning has been booming. It felt like a good time to invest in growing my skill set and learning a little bit more about these new technologies.

There are many data science bootcamps out there — Why did you choose 4Geeks Academy? Was the job guarantee a factor in your decision?

Trevor: The job guarantee was definitely a factor, but what stood out to me was 4Geeks’ strong reputation. I'm someone who does deep research before making any decision, and I went through Reddit threads and online reviews and discussions and everything showed that 4Geeks is a rigorous, challenging program with mentors who are there for you long-term. The lifetime commitment to helping graduates succeed in their careers and being a continuous resource was a huge factor for me. Knowing they stand by their graduates with a job guarantee only reinforced my decision.

Overall, what did the Data Science & Machine Learning Bootcamp curriculum cover at 4Geeks Academy?

Kelsey: We started with the fundamentals, like learning Python and using algorithms. We studied probability, statistics, and linear algebra, and how that all relates to data science. Then we got to dive into exploring databases, learning how to manipulate and visualize data. We learned how to use the basic Python libraries, everything from NumPy to Scikit-learn. 

Finally, we learned predictive modeling. We learned about different types of models, everything from regression to language processing and time series. It was a really good variety! Then we touched on deploying projects before we pulled everything together for our final project.

What is the main problem that your final project is aiming to solve?

Our U.S. Natural Disaster Predictions project is intended to take the most common natural disasters in the United States and predict the annual probability of those disasters occurring. Our hope for this project is to protect communities and better allocate disaster relief resources using these predictions.

How much time were you given to build this project? 

Trevor: We had about 6 weeks. After we established what data set we wanted to use, we hit the ground running.

What are some of the main features of this project?

Kelsey: We ended up using a logistic regression model. We tried a couple of different models and this was the one that had the best accuracy. We have around 60,000 data points that we're working with for this model! When a user enters a year and then chooses from our dropdown list of disasters, it will collate data to let the user know its predictions of the probability of that disaster happening. The data is presented in a heat map over the United States, which we built using JavaScript and D3. The heat map shows which states in that particular year have a higher concentration of tornadoes. If you hover over any individual state, it'll give you an actual risk percentage of the natural disaster occurring within that state. We also have a bar chart listing each state's disaster probability that gives a quick overview of the data. 

Trevor: When we started this project, we noticed that Hurricane Katrina drastically skewed the data set. Since nearly every surrounding county declared an emergency, the events hit an overwhelming spike. To balance the data, we grouped Katrina as a single event rather than counting each declaration separately. Another outlier was the Covid-19 pandemic, which caused an extreme surge in disaster declarations. Since it didn't align with typical natural disaster patterns, we decided to remove it. As we continued analyzing the data, we noticed a long list of disaster types, but many had very few occurrences, so to improve model accuracy, we focused on the top five most frequent disasters. Working within the data set was really fascinating. It became clear that the same data could tell completely different stories depending on how we approached it, and there were endless possibilities for analysis, making it an incredibly engaging challenge.

What were some of the programming languages and tools that your team used to build this project? 

Trevor: Pandas to process the data and Matplotlib to visualize our Exploratory Data Analysis (or EDA). JavaScript to build out the heat map. 

Did you use everything you learned at 4Geeks Academy to build this project or did you both pull in new technologies that you hadn't learned in the bootcamp?

Kelsey: We used everything that we learned at 4Geeks for this project, and then we had to learn some new things on top of that. Since we were working with a multi-label classification problem, we had to learn a lot more about data encoding and how we could manipulate that data to give us the output that we were looking for. 

We had to learn more about using different types of models. We thought that using an LSTM neural network model was going to be our best bet for the data that we were working with. We spent a lot of time learning about that and trying to get a high percentage of accuracy. At the end of the day, we ended up getting the best accuracy with the logistic regression model. Ultimately, it was a cool experience to learn something above and beyond what we learned in the course.

What was your biggest challenge in building this project?

Trevor: I was experiencing the Los Angeles wildfires firsthand while analyzing historical natural disaster data for this project. Outside the air was hazy with smoke and emergency vehicles were driving by, while inside I was focusing on trying to build a tool that could anticipate the very events that were happening outside. My situation made the project feel way more practical. It wasn't just about training a model — it was about understanding how data can help preparedness and response because every air quality alert and road closure served as a reminder of why predictive analysis is very important. This project reinforced why this kind of work matters and not just as an interesting technical challenge. It's something that can provide real world value in helping communities plan for natural disasters more effectively.

Did you demo your project at 4Geeks Academy’s Demo Day?

Kelsey: We had a scheduling conflict so we weren't able to do the big demo day where we invited our mentors and our family. We did a practice demo day and we presented it to the software engineering cohort as well as our classmates and program managers. That was a really fun opportunity to show off what we had been doing!

Trevor, do you anticipate using this project for future tech job interviews?

Absolutely! This project is a great representation of data science skills, from data preprocessing to machine learning and time series forecasting. It showcases my ability to work with real-world, messy data sets, handle imbalanced data, and apply advanced modeling techniques like LSDMs. Plus, natural disasters are a high-impact area where predictive analytics can make a real difference, which makes the project both technical and socially relevant.

In future interviews, I hope to use this project to demonstrate my problem-solving approach. It can show an interviewer how I handle challenges like overfitting and data imbalance, and my ability to communicate complex findings in a clear, actionable way. It's a strong portfolio piece that highlights my technical and analytical abilities. 

What types of tech roles did 4Geeks Academy prepare you for? 

Kesley: I really enjoy the storytelling element that came out of learning how to work with data, so I’m looking at data scientist roles, and potentially data analyst or machine learning engineer roles.

Trevor: I’m mainly looking for data science roles, but I am fascinated with all ends of the spectrum, from data engineering to data analytics.

At this point in your career journey, was 4Geeks Academy worth it for you? 

Kelsey: Absolutely. In particular, having an experienced and knowledgeable mentor is incredibly invaluable. I learned things that I just couldn't have done through independent study.

Trevor: Definitely. 4Geeks Academy was a deep dive into data science. It wasn't just surface-level learning. It was an intensive, fast-paced experience that required a lot of focus and adaptability. Every week we were switching gears and tackling different aspects of data science, from machine learning and deep learning to time series forecasting and NLP. The constant shift forced me to learn how to manage multiple concepts at once, a skill that directly translates to real-world data science work.

What is your advice for incoming data science bootcamp students at 4Geeks Academy? Anything you wish you had known before day one of the bootcamp?

Trevor: My biggest advice for incoming students is to get familiar with coding and Python before starting the bootcamp. You don't need to be a coding expert, but doing some online tutorials will help so that everything isn't completely over your head for the first few weeks. Expect to struggle. Struggling is okay because everyone falls behind, makes mistakes, and feels lost at some point in the bootcamp. The key is to keep pushing forward because things that don't make sense at first will begin to click as you go. Be diligent, put in the effort, and trust the process. Most importantly, don't get discouraged by failure because we all failed multiple times and that's just a part of learning!

Kelsey: I agree with everything that Trevor said. Make sure you have a clear understanding of your motivations for taking the course. Keeping those intentions in mind helps you persevere through the challenging fundamentals. Once you stick with it, everything will come together and it's incredible to see how it all applies to data science.

Find out more and read 4Geeks Academy on Course Report. This article was produced by the Course Report team in partnership with 4Geeks Academy. 4Geeks Academy is a coding bootcamp based in Miami, FL, training in Full Stack Software Development, Data Science and Machine Learning, Cybersecurity, and Applied AI. With 10 locations across the US, they offer part time, online programs as well as hybrid and in person programs.