mrregression is an R package that enables the execution of frequentist and Bayesian linear regression on tall data sets, i.e. n >> d. It can be used with data that does not fit into R's main memory. The implementation follows the Merge and Reduce for statistical models introduced in [1].
[1] Geppert, Leo N., Katja Ickstadt, Alexander Munteanu, and Christian Sohler (2020). Streaming statistical models via Merge & Reduce. In: International Journal of Data Science and Analystics 10.4, 331–347 (2020). https://doi.org/10.1007/s41060-020-00226-0