Skip to content

Advanced operations

The following tutorial will show you how to:

  • upload external data (from your own experiments)
  • preload and leverage data in your new campaigns

Pre-requisites

You should already be familiar with SDLabs and its glossary, and have a premium account to access these features.

Upload external data

Incorporate your existing data seamlessly into experiments by using our platform's upload feature.

Upload external data

1. Access Template Data

  • Go to your Template and click on the "Template Data" button at the bottom of the page.

2. Upload Observations Dialog

  • Click on "Upload CSV" located at the top right of the screen.
  • The upload-data dialog will open. Follow the instructions displayed inside the dialog.

Preload previous campaigns or data

When starting a new campaign, you have the option to preload data from prior campaigns, serving as a foundation for a warm-start approach. You may select as many campaigns or datasets as you like, but keep in mind that preloaded data must belong to the same template (i.e. have identical parameters and measurements), whether from previous campaigns or external sources.

Preload previous campaigns or data

1. Configure Your Template

  • Open a Template with which you already have existing data (be it previous campaigns or uploaded dataset as described above).
  • Configure the template based on your requirements (budget, batch size, optimizer, objective, constraints, parameter ranges).

2. Launch Your Campaign

  • Click "New campaign" using the configured Template.
  • Click the "Preload data" button and select the suitable dataset (e.g. data you have uploaded as in per the "Upload external data" instructions above, or any other relevant campaign).
  • Click "Select" in the preloaded data window and "Launch" in the campaign window.
  • Any observation that was successfully preloaded, and therefore used by the machine learning algorithms to suggest new parameters, will appear in the Campaign page, in the “Preloaded campaigns data” table.
  • Likewise, if some observations could not be used and are ignored by the algorithm, they will appear in the “Logs” table at the bottom of the Campaign page.

3. Visualizing Preloaded Data

  • In the running Campaign, view the "Preloaded campaigns data" table.
  • This table contains data the optimizer will leverage to suggest new parameters.