Actalytics

Inspiration

This project was inspired by our own political work and involvement, as well as the current movements that are occurring. Specifically, one of our team members was a volunteer with the Sunrise Movement, and was part of a campaign to bring renewable energy to Massachusetts. In order to achieve this, they wanted to contact political figures and tell them about the campaign; however, they were unable to find contact information for these figures, and the campaign was not as successful as it could be. In addition, there are numerous movements that are spurring at the moment, and we wanted to create a platform to aid those who are stepping up. Overall, after intensive research and brainstorming, we realized that campaigns don't use the same data driven insights and tools available to startups, we wanted to bring our data science knowledge to create the activism of the 21st century. 

What it does

Actalytics provides key performance metrics and recommendation to guide strategy for campaign organizers. The primary feature of our beta release was ranking the probability of voting on climate action for MA House of Representatives. We came up with a score for each rep, and display the top 5 most likely to engage with a climate campaign. Also, Actalytics tracks key metrics for each outreach channel (email, direct confrontation, print campaign, strike) to provide feedback on the most effective methods of escalation and confrontation. We hope to expand into a full fledged campaign management tool, with contact management, relationship tracking and email bots. 

How we built it

Like most data science, most of the work was data collection and cleaning. We wanted a broad range of features and had to scrape (using scrapy spiders) from a lot of sources. After that, we used pandas and numpy to clean, combine, and format (lengthiest part) into model-ready form. Flask backend with html/css/js created the actual platform, using data from the LogisticRegression model made in sci-kit learn.

Challenges we ran into

We wanted to try collaborative filtering for the recommendation system, but we could not find enough bills with public voting record to get any accurate results. Also, integrating the model into the front end was also a challenge, since we did not do such things in the past. Finally, developing the front end required much thought because we had to think about how the user would approach the page and how they would react, so that we could allow the website to be user-friendly.

Accomplishments that we're proud of

We're proud of learning scraping quickly, as we previously didn't have those skills. Our knowledge of Flask integration also increased. Making the model obtain a relatively high degree of accuracy was quite nice as well, and overall, we were able to acquire many new skills.

What's next for Actalytics

We really hope to expand our model beyond the one use case of climate in Massachusetts House. Going nationally, for all areas of activism, for both the senate and the house will drastically increase our attainable user base. Of course, the platform needs to be filled out as well, but there's a lot more data collecting to be done to expand beyond this Proof of Concept.

Built With

• flask
• html/css/js
• numpy
• pandas
• scikit-learn
• scrapy