Cortex is the easiest machine learning solution on the market. Anyone at any organization can use Cortex to create a predictive machine learning model with just a few clicks of a button – no coding required.
Many of our partners use Cortex to build powerful Event Time Series models, one of several model types that Cortex supports. Event Time Series models train on events that happened in the past and predict whether those events will occur again in the future. Which of your users are likely to become inactive in the next 30 days? Which ones are likely to be upsold? Who will open a promotional email? Event Time Series models can answer these questions and more.
These Event Time Series models allow you to predict how each user will engage broadly with your products and services. But did you know that you can also use Cortex to predict events for specific categories and offerings? Which users are likely to stop using feature X? Which ones may be upsold to package Y? Filtering your positive labels in this way allows you to zoom in on predictions specific to one or more components of your business.
Building Predictive Models
Several of our partners in the eCommerce space use this functionality to build predictive models for category-specific marketing promotions. Say, for example, a new line of pants is set to debut. Cortex can determine exactly which users are likely to purchase pants. Because of this, the retailer can target its promotional campaign effectively.
Filtering your models by category is as easy as checking a box when you specify your positive label within Cortex Models. In this example, Cortex will assign each user a likelihood of purchasing from category Pants within the next 14 days. Without a filter, the model would instead score each user’s likelihood of purchasing at all.
Positive label filtering has valuable applications for our partners across many verticals. Filter your event predictions by product categories, genres, brands, publishers, or however else your data is organized. In addition, you learn more about building category-specific models. To do so, email firstname.lastname@example.org or talk to your account manager today!