Skip to main content

Machine Learning for Social Change

Fernand Pajot (Change.org)
Average rating: ***..
(3.00, 3 ratings)
Slides:   1-PDF 

How do we optimize social change? How do we connect users to the change they want to see?

During this talk, you will learn how Change.org uses machine learning and recommendation algorithms in order to optimize user engagement, revenue, and impact. More specifically, I will cover:

  • how we built one of the most sophisticated machine-learning based email targeting tools ind the industry using finely tuned feature engineering and distributed random forests
  • aggregating multiple data sources (collaborative filtering-based systems, social recommendations, topics a user follows, geographic information) to construct a single feed for a user
  • how we deal with the cold start problem
  • the different collaborative filtering techniques we use and how we scale them
  • how we enrich these techniques with supervised topic generation
Photo of Fernand Pajot

Fernand Pajot

Data Scientist, Change.org

Fernand is the in-house Data Scientist at Change.org, where he is responsible for designing scalable algorithms, data backends and experiments to solve business problems and improve key metrics.