Abstract

A real-world dataset is provided from a pulp-and-paper manufacturingindustry. The dataset comes from a multivariate time series process. The datacontains a rare event of paper break that commonly occurs in the industry. Thedata contains sensor readings at regular time-intervals (x's) and the eventlabel (y). The primary purpose of the data is thought to be building aclassification model for early prediction of the rare event. However, it canalso be used for multivariate time series data exploration and building othersupervised and unsupervised models.

 

Quick Read (beta)