SoMe

SoMe is a Social Media Manager, currently geared toward Twitter users which ofers a varity of different ways to be involved with Twitter. It allows users to schedule tweets and que them for automatic scheduling by custom categories.

This has been my favorite build so far. I really enjoy Natural Language Processing (NLP) most. The original remote cross funtional team of 12 lost two members just a few weeks in. One fullstack developer, and one data scientest. This left me, as a data scientest, with two other data scientists to work with, Amin & Lawrence. We split up the project into two sub projects.

  1. Build and make the model
  2. Figure out how to deploy the model

My primary role was building the model, with Amin. Here’s my personal repo with some of the work I did, exploring and analyzing the data.

Here’s how it works.

A Brief Overview of Twitter

Twitter is a popular social media platform where people can… express their views. A twitter user follows other users and has followers themselves.

What The Model Does

Simply put, it takes a users followers and gathers the tweets they post/interact with. The tweets are then processed and ultimetly returned to the SoMe user, summed up into the most popular topics the followers interact with.

Tweet columns

We can read in (it takes time) up to 5,000 followers and the tweets they engage with using the Twitter API. Thats quite a bit of data to collect and we managed to return results of the top keyword topics.

Conclusion

We still have three weeks to work on this, and we should be able to get pretty far. Most of the Data Science requirements have been met already, but there are still improvements to be made. Check back for updates.