Tennessee-Eastman (TE) is a realistic process simulation of a chemical plant which has been widely used as a benchmark for control and monitoring research. In this Fortran variant of the TE industrial process, model performance is evaluated under 28 types of process faults. These faults occur due to various disturbances; some faults almost instantaneously place the process in a critical state while others produce statistically insignificant deviations.

The development dataset available herein consists of flattened rows of nominal data with dimension 42 and a context of 60 observations (thus 2,520-feature rows). During evaluation, the main program of your submission is expected to access /ingress/te_matlab_revised/testing.csv which has the same format as the development dataset provided. You are then expected to produce a single-column CSV file of anomaly confidence values (in [0, 1]) at /egress/te_matlab_revised/predictions.csv. Your predictions must be sequentially aligned with the rows of the testing dataset.