"Datasets for Multi-scale Decomposition Dynamic Conditional Diffusion Model for Time Series Imputation"

"This study evaluates time-series imputation on six widely used multivariate real-world datasets covering energy systems and environmental measurements: ETTh1&ETTh2, ETTm1&ETTm2, Electricity, Weather, Wind, and Energy. The ETT datasets record electricity transformer temperature and related variables, with two hourly-series and two higher-resolution series. Electricity contains electricity consumption measurements across 321 dimensions. Weather consists of 21 meteorological variables. Wind provides wind-related measurements with 7 variables. Energy includes 28-dimensional energy-related observations. Together, these datasets span diverse temporal resolutions and variable scales, offering a comprehensive benchmark for evaluating imputation methods under different real-world dynamics."