Datasets¶

The library currently hosts 14 datasets spanning 8 domains. Those datasets were collected from various time series imputation works in recent years. The datasets are stored in ./imputegap/datasets and are represented using a numpy.ndarray structure.


Note

The datasets follow this format:

  • Columns are the series/features, separator: empty space

  • Rows are the timestamps’ values, separator: newline

  • Missing values are marked as NaNs


Weather

Air Quality

Water Quality

Neuroscience

IoT Sensing

Sports

Energy

Finance