This week, I spent a lot of time on the proposal presentation. I helped to translate our problem statement into a set of plausible and quantitative user requirements, along with Matt and Thomas. I translated our research document for the proposal into a visual set of slides which matched our team’s brand and vision. Further, I worked on the Music Mirror system diagram, which required a lot of research and thought into how we would actually structure our solution approach. Attached below is the mentioned diagram:
Further, I was the presenter for this proposal presentation, so I spent a lot of time preparing the actual speaking component of the presentation. This included brainstorming a captivating introduction and conclusion, as well as many rehearsal runs.
On the more technical side, I spent time planning out our recommendation system architecture, which revolved around a collaborative-filtering design. Further, I found good datasets which we could use to train and validate our model, including the last.fm dataset (http://millionsongdataset.com/lastfm/) which is a great candidate for our purposes. Lastly, I began to get familiar with the Spotify API structure.
We are on schedule with the project so far. Next week, however, there are quite a few things we aim to complete. Especially, we will begin building our project codebase starting with the interactions with the Spotify API
Next week, I plan to make a final decision about how to proceed with the ML model versus writing a wrapper system around the Spotify API recommendation features. With that, I will have functional preliminary progress on the recommendation portion of our product. Further, I’d like to test around with the Spotify API music player to figure out how exactly we will be transmitting song resources to the Bluetooth speaker system. With that being said, we will work to clearly define our system’s solution design in preparation for the design presentation and document.
