A small library for simplifying the training process of a deep learning time series classification task using a PyTorch backend, and provide prediction explanations using the Captum library capabilities.
You can train, evaluate and deploy deep learning models using a simple sklearn-like interface with the classifiers provided in this section:
- MLPClassifier
- CNNClassifier
- GRUClassifier
- LSTMClassifier
Currently, all classifiers only support the Pandas DataFrame format for time series data.
A small dataset class used for converting DataFrames into Torch Datasets.
Using TSExplainer's .interval_plot() you can see which time intervals contributed the most for your model final prediction, with support for multi-dimensional series.
