tsfresh leads in time-series feature set comparison study

A new study published on arXiv compares six open-source time-series feature sets and three baseline sets across 124 classification tasks. The research found that while feature sets performed similarly overall, with 85.3% of pairwise comparisons resulting in ties, the tsfresh library showed the strongest performance with 29.03% wins against other sets. The study highlights that the specific composition of a feature set can significantly impact classification performance, and simple baselines using Fourier coefficients and quantiles can be effective for certain problems. AI

IMPACT Provides insights into the effectiveness of different time-series feature extraction methods for classification tasks.

RANK_REASON The cluster contains a research paper published on arXiv detailing statistical comparisons of time-series feature sets. [lever_c_demoted from research: ic=1 ai=1.0]

Read on arXiv stat.ML →

• alphaXiv
• arXiv
• CatalyzeX Code Finder for Papers
• CORE Recommender
• DagsHub
• Fourier coefficients of modular forms on G2
• Gotit.pub
• Hugging Face
• Quantiles
• ScienceCast
• tsfresh

• paper
• other

AI-generated summary · Google Gemini · from 1 sources. How we write summaries →