{"references": ["Taieb, S.B., Bontempi, G., Atiya, A.F., Sorjamaa, A., 2012. A review and comparison of strategies for multi-step ahead time series forecasting based on the nn5 forecasting competition. Expert Systems with Applications 39(8), 7067 - 7083.", "Neural Forecasting Competitions, 2008. NN5 forecasting competition for artificial neural networks and computational intelligence. Accessed: 2020-05-10. URL http://www.neural-forecasting-competition.com/NN5/"]}

This is the aggregated version of the daily dataset used in the NN5 forecasting competition. It contains 111 weekly time series from the banking domain. The goal is predicting the weekly cash withdrawals from ATMs in UK. The original dataset contains missing values. A missing value on a particular day is replaced by the median across all the same days of the week along the whole series and then, have been aggregated into weekly.