Sensus

Inspiration

One of the biggest challenges that small businesses face is finding a good market to fit into. It is the reason a lot of them are not able to compete in a saturated market with other strong competitors already in the pool. In some cases, these small shops can help certain communities gain access to the goods they need or provide more accessibility for community mobility. Unfortunately, this would be hindered through poor access to markets for small, local businesses. However, our tool aims to provide a simple, elegant, and novel solution to this issue.

What it does

Our application, Sensus, allows small business owners to find product-market fit through analyzing markets via tweets from Twitter. The user is able to enter a keyword query alongside specific locations they want to understand the market of. This information is fed into a model trained on tweets as data to conduct sentiment analysis to gauge the general sentiment of that keyword or product in the location specified by the user. Through Sensus, small business owners can understand where to target their marketing, their future growth, and potential future customers through analyzing these niche markets.

How we built it

We primarily used Streamlit, a Python library, to create the user interface for the project. We used Tweepy to aggregate the tweets within the area of interest and Naive Bayes to analyze sentiment. The specific parameters that were analyzed were keyword (query/product), location (lat, long, cit, state, zip, radius), number of retweets, and the number of likes.

Challenges we ran into

We had trouble with integrating a map into our interface for the user to drop a marker where they want their search to be centralized. Furthermore, since the new version of Tweepy does not include location functions built-in, we had to develop our own location pruning algorithm that selects tweets from the location specified by the user to conduct the sentiment analysis.

Accomplishments that we're proud of

One of the major things we were proud of from this project was implementing a simple and elegant user interface that can easily be understood and used. In addition to the key design aspects of the application, the underlying algorithm produces an understandable and intuitive output to the user that can be used to determine whether their product would be popular in a specific location via sentiment analysis. With these two factors combined, we were impressed by the potential impact this could have on small businesses getting their feet on the ground, running, with the products and ideas.

What we learned

Through this process, not only did we learn the value of team collaboration and ideation, but we were able to harness and strengthen technical skills. For instance, we learned about a new system of creating web applications through Streamlit. Additionally, going through design iterations of the Naive-Bayes model was a huge, but satisfying, learning curve. Furthermore, developing the location pruning algorithm was an interesting challenge that reminded us of the importance of time complexity and efficiency of algorithms, which would have been overlooked if a package already existed for it.

What's next for Sensus

The next step for Sensus is clear: deploy the application. The amazing thing about Sensus is that it is a production-level application that can aid small business owners to find their niche markets to target, therefore, helping them find product-market fit: a key issue in the space of businesses. However, the potential of growth for Sensus does not stop there! With a few tweaks, Sensus could be an application that conducts sentiment analysis on social media apps regarding politics and inform governments about current political climates. Additionally, specific to the current application of Sensus, we can move onto the second level which is to actually bridge the connection between users and small business owners.