How it work´s Features known in time series model "retail"?

[fernandomartin]

We´re a bussiness retail, the case is next:

We would like to predict the beer category, training data are:

• Daily date (01 junuary 2019 - 30 may 2021).
• Inventory (pieces or units)
• Sale price per unit.
• Daily sales. 
• "id ITEM" (10 products).
• average sale per store
• Days of inventory.

We would like to predict 01 june 2021 - 30 september 2021 per product, per day.

The question are two:

1.- How impact´s if i set a "Features known" at training time, it´s modify the forecast like a "what if"? 

2.- How to create different scenarios if my price change? Example: "Corona beer" sales price is $52, per piece, if we change it to $46 now, which is te new forecast?

thanks for you support and help friends.

[jacob_a]

the datarobot pathfinder site has lots of different use case, some for time series modeling too

like https://pathfinder.datarobot.com/use-case/forecast-demand-new-products-aka-cold-start-modeling?tab=o... https://pathfinder.datarobot.com/use-case/forecast-daily-demand-store?tab=tech

and theres this blog https://community.datarobot.com/t5/resources/demand-forecast-single-series/ta-p/4884 (looks a little older but I think it's still good)

good luck!