A user-defined data type for the storage of time series data allowing efficient similarity screening

Area of Science:

• Computer Science
• Bioinformatics
• Database Systems

Background:

• Time series data volume is rapidly increasing.
• Existing storage solutions (arrays, blobs) are inadequate for efficient time series data management.
• Current indexing methods lack integration into robust database systems.

Purpose of the Study:

• To develop a specialized data type for time series storage within PostgreSQL.
• To integrate an efficient access method for time series data into a database system.
• To address the limitations of current data storage solutions for large-scale time series datasets.

Main Methods:

• Developed a novel time series data type for PostgreSQL.
• Implemented an access method utilizing Symbolic Aggregate approXimation (SAX).
• Tested the data type and access method on real-world (plant gene expression) and simulated time series data.

Main Results:

• Successfully integrated a new time series data type into PostgreSQL.
• Demonstrated efficient data storage and retrieval capabilities.
• Validated performance on both large-scale experimental and simulated datasets.

Conclusions:

• The developed PostgreSQL data type provides an effective solution for managing large volumes of time series data.
• Integration of SAX-based access methods enhances database performance for time series analysis.
• This approach offers a promising direction for future database system development in handling scientific data.