Data modeling low-cardinality data

In a traditional database, low cardinality data is typically normalized into separate tables and keys are used to maintain the relationships.

This FAQ guides you through:

• selecting the best unique identifier for each row of data
• CSV structure
• SQL statements to create a target table and read data from the CSV file
• querying the data

Table of contents

• Before you begin
• Sample data
• Which unique identifier should I choose?
• Next step

Before you begin

• Learn about data cardinality

Sample data

Which unique identifier should I choose?

The first column in every FeatureBase table is the unique identifier _id and can be assigned one of two data types:

• ID data type
• STRING data type

A CREATE TABLE statement with _id STRING will increase disk requirements and query duration.

• Learn how data is encoded by FeatureBase
• CREATE TABLE statement

The choice of unique identifier helps reduce the number of rows to be ingested, which then increases the speed of your queries.

In this example, Vertebrae is the best choice for unique identifier because it reduces four rows to two:

Next step

• CREATE TABLE myspecies

---