This package contains helpful functions for preparing and understanding data output to be used in different Keras architectures.
!!Important: Currently only tools used for RNNs have been implemented for this project. Additional features described in the roadmap below are planned for addition. If you have suggestions please open an Issue and flag it as an enhancement.
Full documentation can be found here.
- Finish predict_ts api
- Build predict_ts tests
- Create unscale api
- Build unscale tests
- Run through tests again and add where coverage is missing
- Update docs with full comments and examples
- Refactor readme page
- Submit to pypi
- Universal prep
- train_test_split: Function to create training and test data from different data sets
- scale: minmax, standard, log, boxcox transformations to the data. Fit and transform train, transform test and validation only.
- RNNs
- transform_for_rnn: Create time series data set (np arrays)
- CNNs
- transform_for_cnn: Create np arrays
- Pre Model
- get_input_shape: return input shape from training data for model
- Post Model
- unscale: reverse scaling methods. User passes new array, gets back
- ts_predictions: handle ts steps to predict for future
- model_summary: Output model arch summary and
Web and video tutorials are available to those who are willing to become a sponsor. Visit my sponsor page to learn more.