The problem of imbalanced data is common in real-world domains with several solutions being put forward in the last decades, especially for classification tasks. More recently, this problem has also started to be considered for other predictive tasks such as regression. Resampling solutions are a popular and effective way to deal with this problem. However, their computational requirements make them impractical and/or unsuitable for larger datasets. So far, only one attempt has been made to create a distributed version of a resampling algorithm for regression. The existing DistSMOGN algorithm uses clustering to optimize SMOGN’s computations. In this paper, we propose DistResampleR-Lite, a new flexible framework for efficient distributed resampling in regression tasks that combines an accelerated K-Means clustering algorithm to partition the dataset with the approximate calculation of nearest neighbors. DistResampleR-Lite is also flexible and can be used with any resampling technique. Our experiments with different resamplers show the advantages of DistResampleR-Lite in terms of time and performance across multiple datasets.