Brainstorm your Next Data Science Project

This article will be a guide for how to brainstorm for your next data science project. At the same time, I will provide my thought process on how I am brainstorming for my data science capstone. I’m approaching the end of my Data Science Immersive program at General Assembly, and it’s time for me to show off what I know.

Questions to ask yourself when planning your next Data Science project:

What did/do you enjoy learning?

Think about all of the subjects you enjoyed learning. In the past, what subjects excited you and made you want to practice more? What data science topics made you think about future projects? Is there a particular project you want to expand on — adding stretch goals?

Some of the lessons I enjoyed during my Data Science Immersive was NLP(Natural Language Process) and Web scraping. I see myself working with a lot of web scraping for future projects. Web scraping is tedious but I want to incorporate this in my capstone to give myself practice for future projects. NLP is fun for me because it allowed me to use my qualitative skills. I enjoy investigating text data from large pools with many subjects (ex. Reddit and Twitter).

What do you want to learn more about?

What data science subjects do you want to improve on? Has there been a data science practice that you haven’t gotten a chance to work with yet? A capstone project is a perfect time to experiment with any subject you want to learn more about because you have experts by your side. Especially at General Assembly, all of the Data Science Instructors’ are the most intelligent people I’ve interacted with. They are great resources to help guide you through your next project.

I haven’t gotten much a chance to work unsupervised learning. The week we learned unsupervised learning I was very excited. It is definitely a weird subject because there is no “goal” in the models — similar to marketing and politics! Unsupervised learning is used to find trends. In a real-world politics and marketing project, once these trends are identified, then a plan can be created to determine how to target specific clusters. This is what intrigues me.

What are you passionate about?

This is probably the most important aspect to consider in your capstone. It’s important to work on a project you care about. Don’t work on a sports-related project if you have no interest in sports. Don’t work with social media if you don’t understand it. You want to work with a subject you have an interest in where you can bring in your own expertise and perspectives.

When brainstorming for a capstone, the great data science instructors at General Assembly advised our class to think about what we are passionate about. This is hard for me. All I’ve done in the past 6 years was work, go to school, and run — boring. I’m not much of Netflix binger, video game player, and don’t really understand trends on what I like to call klout-related apps (Instagram, tik-tok, snapchat, etc.). However, one app I pay loyalty to is Twitter aka, my morning news-paper. Twitter is where I catch up with all of my news of interest. My daily feed consists of track and field, LA-related sports, politics/news stories, and puns.

Are you up for a challenge? What skillsets do you plan to show off with this?

Think about your goal for this project. A capstone project is something that future employers and fellow data scientists will ask you the most about. Working with techniques you feel comfortable with is good because it gives you the space to specialize in a particular subject. On the other hand, you may want to challenge yourself by working with techniques you are unfamiliar with. There is no wrong answer. Think about what is best for you.

My Top Data Science Ideas:

1. Passion related project: Create a calculator to predict how a fast a runner will be in a specific track and field event for their upcoming season. This would include scraping data from several track and field websites. I would have to include time-series analysis in my model.
2. Comfort related project: Build a new-artist recommendation system based on song lyrics. This was inspired by another Medium article I read in the past. This project would allow me to stay in my comfort zone and work with NLP.
3. Challenge related project: Attempt to reclassify NBA players based on their playing style. This will be a challenge because I will have to practice web scraping, unsupervised learning, and working with numerical data. Web scraping is a challenge because I will have to learn the HTML formats of multiple sites. I do not have much experience working with unsupervised learning and compared to NLP, analyzing numerical data is more challenging for me.

Things to Consider:

1. It’s not all about numbers — Don’t focus on what projects would give you the best accuracy score. Some of the most creative projects don’t have the best results; however, they show great technique through their data gathering and EDA process.
2. Take advantage of your resources — If you’re in a data science immersive like myself, your instructors are there to help. USE THEM. They are here to help. They can help guide your thought process and point you in the best direction.
3. Schedule yourself — We all know data science projects take time and there is always something that delays the process. Once you choose a project, spend a day creating a plan on how you want to complete everything. Use planning tools such as Trello or Wunderlist. If you’re a social media addict, download Moment on iOS or Offtime on GooglePlay, to ensure you won’t get distracted by certain apps during project work-time.