Cortex automatically connects every step of the Machine Learning process into end-to-end Machine Learning Pipelines that anyone in your business can run. In this guide, we’ll discuss the fifth and final step of a Cortex pipeline: prediction.
What is prediction?
The prediction step of your pipeline comprises two processes:
- Building features for all objects (e.g. users) that will get a prediction, and
- Feeding those features through the pipeline’s winning model.
Together, these two actions result in a prediction being made for each object.
Below are a few notes that are important to understand about the prediction step.
- During model selection, your pipeline learned a relationship between a certain set of features and a particular goal. So, the features that are built during the prediction step will match those selected during the feature engineering step. The only difference will be either which objects the features are built for (Classification, Look Alike, Regression) or the time period over which they’re built (Future Events).
- Your predictions will take different forms depending on which type of pipeline you’ve built
- A Future Events pipeline outputs a conversion probability for each user
- An Uplift pipeline outputs a score between 0-1 for each user, where a higher score means that your intervention is predicted to have a larger impact on that user’s likelihood of converting
- A Regression pipeline outputs a continuous value for each user
- Look Alike and Classification pipelines output a score between 0-1 which represents how similar each user is to your positive labels
- A Recommendations pipeline outputs a ranked list of items for each user, along with a score for each item indicating Cortex’s certainty that the user will interact with that item next. Scores across all item values sum to 1 for each user.
Why does it matter?
Predictions are the final output of your end-to-end pipeline. Deploying these predictions into automated workflows ensures that your ML Pipelines are always working to optimize your business.
- What is a Machine Learning Pipeline?
- Data Preprocessing
- Data Cleaning
- Feature Engineering
- Model Selection
- Accessing Predictions
Still have questions? Reach out to firstname.lastname@example.org for more info!