To structure the output of your experiments with separate columns for results and metadata, you can utilize the `phoenix.Schema` to define your dataset schema, specifying distinct columns for each type of data. According to the <https://docs.arize.com/phoenix/~/changes/v6Zhm276x8LlKmwqElIA/how-to/define-your-schema|Phoenix documentation>, you can define a schema with specific column names for predictions, actuals, features, and tags. This allows you to organize your data in a tabular format with clear distinctions between different types of information.
If you want to add a new column to store additional metadata, you can modify your dataframe before uploading it to Phoenix. For example, you can add a new column to your pandas dataframe and then define it in your schema as a feature or tag column. This approach ensures that your results and metadata are stored separately and can be easily accessed and analyzed.
Would you like more detail on how to implement this in your code?
Sources:
- <https://docs.arize.com/phoenix/~/changes/v6Zhm276x8LlKmwqElIA/how-to/define-your-schema|Phoenix documentation>
