Added Changed Fixed - fixed the default hidden_layers_type arg in `_forecast_rnn()`
0.6.8
Added Changed Fixed - fixed the rnn estimator which was using a dict to set hidden layers without unique keys, causing all layers except the last to be ignored, so all rnns were returning one layer results. Now the args in that function are different, but tested and it works to add more layers.
0.6.7
Added - added `N_actuals` to `__repr__()` function Changed - Took pt normalizer out of all example grids since it still errors out sometimes without being obvious why - `notebook.results_vis()` no longer prints anything and automatically sets the plot title instead - `Forecaster.set_test_length()` now accepts fractional splits Fixed
0.6.6
Added - added the `add_sklearn_estimator()` function so a user can add any sklearn regression model to forecast with. Changed Fixed - some of the examples weren't plotting correctly after recent updates.
0.6.4
Added - added residuals to the function `export_fitted_vals()`. Now gives dates, actuals, fitted vals, and residuals - added `multiseries.export_model_summaries()` function Changed Fixed - `notebook.results_vis()` broke after last update and has now been fixed
0.6.2
Added Changed - the `plot()`, `plot_test_set()`, and `plot_fitted()` functions now return figures instead of automatically plotting for you. This means more customization is now possible. Took out `to_png` and related args from these functions since that can be done now with matplotlib Fixed - changed the insample metric evaluations for RNN and LSTM to be on the full training set instead of just the last few observations.