TerseTS: A Framework for Time Series Compression
TerseTS is an open-source framework for lossless and lossy time series compression. TerseTS is written in Zig and unifies over a 0.08 0.06 dozen compression methods under one API with C, Rust, Julia, and Python bindings, enabling the consistent use and comparison of these methods. An extensible architecture accommodates diverse algorithmic paradigms and exposes encoding and decoding primitives, enabling the creation of custom compression pipelines. Our demonstration has an interactive dashboard that 0.04 0.02 Decompression RMSE 0.00 0 50 Saugeenday 100 Compression Ratio 150 0 50 100 Oikolab 150 Compression Ratio supports data upload, visualization of reconstructed time series, and comparison of compression trade-offs across different methods. Participants can also design and evaluate custom compression pipelines and compare those with state-of-the-art methods.
