Machine learning is a form of artificial intelligence that involves applying techniques such as pattern matching and computational statistics to large amounts of data to predict events. Simply put, it’s training “machines to learn” -- hence the name. In Lytics’ case, we apply machine learning to companies’ customer data to help make predictions around what people may do next: make a purchase, leave the brand, have affinity for a certain type of content, go on a binge buying spree, and so on. This is how we uncover highly valuable audience segments such as “Likely to Buy” that you can narrowly target in online advertising, email marketing and web personalization campaigns.
What makes our machine learning unique from that of other customer data platforms is we look at more than 125 factors about a person’s behavior to predict what they’ll do next. Other companies look at only a handful of aspects of people’s behavior. More technically speaking, Lytics implements an ensemble of statistical techniques to describe a person across a set of linearly independent behavioral dimensions. Techniques include random forests, logistically weighted moving averages and quantile estimation.