Data Preparation

Supported formats & requirements

CSV requirements

  • First row must be column headers.
  • One experiment per row.
  • Missing values: empty cells or a consistent marker such as NA.
  • Column headers should be unique — duplicate names are a common silent failure.
  • Excel files are also accepted.

What column types exist

  • Continuous — measured quantities like temperature, pressure, conversion.
  • Categorical — discrete labels like material family, composition/matrix.
  • Discrete — counts and integers.
  • Boolean — true/false flags.

Size limits

Maximum file size and rows per dataset depend on your tier (see Pricing). Files above the limit are rejected with a clear message — uploads are validated before any compute starts.

Prefer . as the decimal separator. Commas inside numeric values (e.g. "1,234.5") confuse the parser — strip thousands separators before uploading.
← Previous
Active Learning — closing the loop
Next →
Column roles & common mistakes
← All docsReport a problem on this page →

Go from reading to running

The quickstart tutorial walks the first synthesis, evaluation and prediction on a demo dataset — each step matches a real platform page.