Inspiration
Natural language processing is one of the most rapidly evolving fields in computing today. Small talk about the weather is one of the most mundane things a person does each day. What can we learn about people's moods from analyzing online weather conversations?
What it does
Essentially, WS connects daily weather data with data about the aggregate sentiment of weather tweets for that day. Using regression analysis, the app attempts to find patterns in how people talk about the weather and what the weather pattern is. It also attempts to make predictions about how people will feel about possible future weather scenarios.
How I built it
Through a lot of research, I found the Twitter developer API, the NOAA daily weather data, Tweepy python library, and Textblob python library for sentiment analysis.
Challenges I ran into
It was difficult to find daily weather data over many days from a commercial vendor (most supply only the last few days), but eventually I was able to turn to the federal government for good data. However, the Twitter API only lets free users go back one week, so I ended up with far less data than is really necessary for a machine learning project.
Accomplishments that I'm proud of
I arrived at HackUNO deciding to do something with machine learning, but I didn't really have an idea of what that meant. Over the course of the 24 hours, I worked with many new tools I had just learned about, I learned about data science methods and practices, and got to start something really cool.
What I learned
It's always best to plan the project ahead of time. One of the main reasons for this is that challenges will come up, and your project can be significantly handicapped halfway through in a way that may have been foreseeable. That being said, it's difficult to really do anything new in 24 hours, because there is so much you can't really plan for.
What's next for Weather Sentiments
Unfortunately, probably nothing due to the lack of Twitter data available for my purposes. However, I will definitely be using the concepts and tools that I worked with again.
Log in or sign up for Devpost to join the conversation.