The official PyTorch implementation of the paper "SAITS: Self-Attention-based Imputation for Time Series". A fast and state-of-the-art (SOTA) deep-learning neural network model for efficient time-series imputation (impute multivariate incomplete time series containing NaN missing data/values with machine learning). https://arxiv.org/abs/2202.08516
machine-learning deep-learning time-series interpolation pytorch transformer imputation attention attention-mechanism irregular-sampling incomplete-data imputation-model missing-values self-attention impute time-series-imputation partially-observed-data partially-observed partially-observed-time-series incomplete-time-series
-
Updated
May 28, 2025 - Python