Skip to content

FAQ

Noon is an open-source project that builds tools for data analysts. We focus on the gap between raw data and reliable analysis — helping you understand, validate, and transform your data with confidence.

Our first tool is FineType, a semantic type classifier for text data.

FineType detects what kind of data is in your text columns. Point it at a CSV and it tells you which columns contain dates, emails, IP addresses, coordinates, phone numbers, and 160+ other types — with confidence scores.

Each detected type maps to a DuckDB SQL expression that will parse the data correctly. Profile first, then cast with confidence.

FineType achieves 92% accuracy on its test set. On real-world data from the GitTables benchmark (2,363 annotated columns from 883 CSV tables):

  • Timestamps: 100%
  • Country names: 100%
  • URLs: 89.7%
  • Dates: 88.2%
  • Person names: 80–85%

Column-mode inference improves accuracy on ambiguous types like dates and coordinates by analysing the distribution of values across the column.

FineType profiles CSV files directly. You can also classify individual values or a column of values from a text file.

For querying, Noon tools work with DuckDB, which natively reads CSV, JSON, Parquet, Excel, and many more formats.

Yes. All Noon tools are open source under the MIT licence. Free to use, modify, and distribute.

File issues or open pull requests on GitHub: