Beta 0.5.0rc1

[ourownstory]

Relevant new features:

• [lightning] GPU support #961 GPU (and other accelerator) support
• [feature] Plot components of averaged df (and quants) if multiple IDs, but no df_name specified in plot_components() #888 Plot multiple sets of parameters due to Glocal model as mean and 10/90 percentiles.
• [feature] separate panel uncertainty plot #779 New panel in plot_forecast depicting uncertainty
• [feature] Support plotting of only specific panels in plot_parameters and plot_components #884 Plot select panels in plot_parameters and plot_components

API and modeling changes:

• Lightning Migration #837 Migrate backend to PyTorch Lightning
• Remove support for df dict as an input #929 remove support for dictionaries as inputs. Please use a single DataFrame with an ID column for each time series
• change deprecation message to switch plotting backend to plotly #984 deprecation warning: upcoming change of plotting backend to plotly (matplotlib support will be dropped in future)
• Remove residuals in forecast dataframe returned by m.predict() #927 remove addition of residuals to forecast dataframe and subsequent plotting

Important fixes and enhancements:

• [bug] Forecast output dataframe contains all float dtypes instead of having some objects #978 Fix None dtypes to NAN
• Resolved discontinuous trend when changepoints are not sorted chronologically #968 Sort changepoints
• Add alias function plot_last_forecast() for plot_latest_forecast() and give deprecation warning #919 backwards compatibility for plot_latest_forecast
• [typing] Add type annotations for main NeuralProphet class #981 Add type annotations for NeuralProphet class
• [feature] Support setting custom quantile in plot_parameters() #853 new argument for setting custom quantile in plot_parameters
• Mask seasonality on Sat/Sun when data frequency is Business days #808 Do not plot weekend seasonality for business day data frequency

Additionally, many smaller bugfixes and improvements to the codebase were also introduced.
For details, please view the merged Pull Requests.

---

This discussion was created from the release Beta 0.5.0rc1.