Khiops Sample Datasets

These are publicly available datasets reformatted to easily test the Khiops AutoML suite.

Each dataset directory contains:

• One or more TSV files with the data tables
• A Khiops dictionary file .kdic describing the table schema
• Optional supplementary information and scripts

Datasets

Adult

• Source: http://archive.ics.uci.edu/dataset/2/adult
• Description: Also known as "Census Income" dataset. Main task: Predict whether income exceeds $50K/yr based on census data.
• Dimensions: 48842 instances, 14 variables.

Iris

• Source: http://archive.ics.uci.edu/dataset/53/iris
• Description: The data set contains 3 classes of 50 instances each, where each class refers to a type of iris plant.
• Dimensions: 150 instances, 5 variables.

Letter

• Source: https://archive.ics.uci.edu/dataset/59/letter+recognition
• Description: Database of character letter image features. Main task: Identify the letter.
• Dimensions: 20000 instances, 17 variables.

Mushroom

• Source: http://archive.ics.uci.edu/dataset/73/mushroom
• Physical characteristics characteristics from various mushrooms, from the Audobon Society Field Guide. Main task: Predict if a mushroom is poisonous or edible.
• Dimension: 8124 instances, 22 variables.

SpliceJunction

• Source: https://archive.ics.uci.edu/dataset/69/molecular+biology+splice+junction+gene+sequences
• Primate splice-junction gene sequences (DNA) with associated imperfect domain theory. Main task: predict the type of splice junction. Multi-table database.
• Dimensions: (after removal of duplicates among the 3190 initial instances)
  • Main entity SpliceJunction: 3178 instances, 2 variables.
  • Secondary entity SpliceJunctionDNA: 190680 instances, 2 variables.

Accidents

• Source: https://www.onisr.securite-routiere.gouv.fr/
• Description: From Observatoire national interministériel de la sécurité routière (ONISR), a traffic accident database from open data. Multi-table database.
• Multi-table database with 4 tables in a snowflake schema, 57783 instances in the root table
• Dimensions:
  • Main entity Accidents: 57783 instances, 13 variables.
  • Secondary entities:
    • Places: 57783 instances, 18 variables.
    • Vehicles: 98876 instances, 9 variables.
      • Users: 130169 instances, 16 variables.

AccidentsSummary

• Simpler version of Accidents, containing only the Accidents and Vehicles tables and less columns.

Customer

• Tiny artificial dataset. It illustrates a snowflake schema.

CustomerExtended

• Tiny artificial dataset. It illustrates a snowflake schema with external tables.

Reference

• Bache, K. & Lichman, M. (2013). UCI Machine Learning Repository http://archive.ics.uci.edu. Irvine, CA: University of California, School of Information and Computer Science.