Keras & Sklearn interface to leverage Neural Nets for forecasting

[David L]

Hi everyone, 

I’ve been using both Keras and Sklearn for a while now and realized that their NNs could be used to do very accurate forecasts which could then be used by supplychainpy.   I guess this question is for Kevin as how one might use Sklearn for the forecast and then just upload/reference it in his packages.  

[Kevin Fasusi]

Hi David,

Thanks for getting in contact, your work with NNs for forecasting sounds interesting. Unfortunately, there isn't a way to upload or reference an alternative forecast. 

It would be awesome to both support NNs inside the library and offer the facility to reference an external forecast. To explore the possibility of referencing an external forecast, it would be useful if you could share dummy output data showing the structure and data points for your NN forecasting model. With the structured dummy output data, we can explore the possible data structures to use and how to reference them in the library in a more generic way.

If you are up for moving this forward, please raise a ticket on GitHub for 'Referencing External Forecast' and add the 'new feature' tag. If you can, post the dummy data as mentioned previously, and we can explore a possible solution.