Benefitter: Early Classification on Multivariate Time Series

Requirements

For this code to run properly, the following python packages should be installed:

numpy
scipy
sklearn
Keras
keras-self-attention [Optional]

To run experiments on the UCR dataset, please download the datasets from http://www.cs.ucr.edu/~eamonn/time_series_data/ and copy it to data/ucr while preserving the subfolder structure.

To run the code, use the following command:

python keras_clssification/benefit_predictor.py --data_path "datasets/ucr" --dataset_name [name] --epochs 1

here [name] = ucr_dataset_name e.g. ECG200.