This post features two projects from recent graduates of our data science bootcamp. Take a look at what's possible to create in just 12 weeks.
Clustering NBA Playstyles Using Machine Learning
James Fan, Metis Bootcamp Graduate
Metis graduate James Fan loves everything about basketball, so it made perfect sense for him to fuze that passion with data science for his Project #4 (of 5) during the bootcamp. Based on one particular blockbuster trade that happened over the summer (Kevin Durant to the Nets and D'Angelo Russell to the Warriors), he wanted to answer the question: Can we use machine learning to place NBA players into categories to predict how a player fits in on a given team?
In a blog post about the project, he explained further: "The goal of the project is to determine the types of players and their roles based on their activity or the space they use. (See full list of these features within the post.) Stats such as points, rebounds, assists, steals, blocks, etc. were NOT included as features as they are dependent on data like minutes played (also not included) or number of shots. Including stats like points, rebounds and assists allows for the possibility for the results to be largely based on those features, which is not the goal."
Want to read more about this project? Get much more detail here.
Automatic Pricing for Etsy Sellers
Asmita Kulkarni, Metis Bootcamp Graduate
As some readers will already know, Etsy is an online marketplace where creators sell millions of handmade goods. For her final project, bootcamp graduate Asmita Kulkarni was interested in the idea that "many of these sellers are new to the website and may not be aware of the market or their competitors and pricing techniques for their products," she wrote in a blog post about the project.
With that in mind, she set out to use machine learning to create a tool that provides price estimates on sellers’ products.
"I want to provide a price estimate to the sellers whenever they upload a new product to their shop. To do so, first, I want to find “similar” products (using k-prototypes algorithm for clustering products together). Then, I want to use this “cluster” label as a feature, along with other attributes, in a Linear Regression algorithm to predict the price of the item," she further explains in her post.
Read it in full here to get far more detail on how it all turned out and how she got there.
See more examples of Metis student projects here.