Predictive Coding


Predictive coding uses machine learning to predict which documents in a case might be relevant. It's most often used to prioritize review or identify documents that can be ignored. Learn how to get started by creating your model.

After you’ve created your predictive coding model, it’s time to interpret your results! You can access your predictive coding models from the Document Analytics menu. The predictive coding model page is separated into three sections: action items, results, and training.

There are multiple ways to improve the accuracy of your predictive coding models.

Interactive Training

Live Training

Everlaw offers free live and recorded training sessions on Predictive Coding which you can sign up for here.

Topics covered include:

  • How to set up assignment groups and distribute documents for review

  • How to create a predictive coding model and use that model to find documents and organize a document population

  • How to use the predictive coding and assignment tools together on Everlaw to prioritize document review by incorporating the output of a predictive coding model in an assignment workflow

Training Activities Checklist

After you've watched videos and have a basic understanding of how to use Predictive Coding, practice your skills and try performing the following actions:

  • Create a new predictive coding model and review at least 200 documents with 50 relevant and 50 irrelevant

  • Use the coverage graph to select documents for review to improve your model

  • Create weighted terms to help train your model

  • Create a training set from a randomized set of documents

Next: Learn about Search Term Reports