Upload a Dataset
Use historical data to warm-start your optimization. There are two ways to incorporate previous experimental data.
Option 1: Start an Experiment From a Dataset
You can create a new experiment by uploading a CSV file as the starting point. The platform will automatically define the variables and results based on the columns in your file.
How it works:
Upload your CSV file during experiment creation.
The platform detects the columns and creates the corresponding variables and results.
You can expand the range of numerical variables (e.g. widen bounds), but you cannot add new options to categorical variables at this time.
Your historical data is automatically included as the starting dataset for the optimization.
Option 2: Upload Historical Data After Experiment Setup
If your experiment is already set up and running, you can upload historical data as a CSV to replace the current suggestions with your own data points and submit them.
How it works:
Download the current suggestions as a CSV to see the expected format.
Format your historical data in the same CSV structure (same column names and order).
Upload the CSV — it replaces the current suggestions with your historical data.
Submit the results as you would with normal suggestions.
This approach is useful when you want to feed prior experimental results into an already-configured experiment so the optimizer can learn from them.
Good to Know
Historical data helps the optimizer make better suggestions from the first iteration.
Supported formats: CSV (.csv) and TXT (.txt). Excel files are not supported for upload.
When starting from a dataset, the experiment setup is limited to the variables and results defined in your file.
When uploading after setup, use the same CSV format as the downloadable suggestions template.