Timeline for Forecast time series data with external variables
Current License: CC BY-SA 3.0
7 events
| when toggle format | what | by | license | comment | |
|---|---|---|---|---|---|
| Jan 11, 2021 at 12:51 | comment | added | Stephan Kolassa | We may not have put it quite this way, but we discuss these three types of causal variables in Fildes et al. (in press, IJF). | |
| Jan 11, 2021 at 11:54 | comment | added | Numbermind | Yes! But I was asking for an actual reference or github for example! But thanks regardless! | |
| Jan 5, 2021 at 11:45 | comment | added | Stephan Kolassa | Well, it happens in my day job: forecasting supermarket sales. Features include calendar events and seasonality, and we know when Christmas happens and what time of year it is. Other features include promotions and prices, and these are set by the retailer. Yet other features may include the weather, and here we need weather forecasts. | |
| Jan 5, 2021 at 11:21 | comment | added | Numbermind | Can you give me a link or a reference where that happens? I can only find material where ALL of the variables are forecasted together. I'd like to only forecast target variable y while using "future" feature variables. | |
| Jan 5, 2021 at 8:09 | comment | added | Stephan Kolassa | @AmateurMathematician: yes, that is precisely what I am saying. You need future values for your explanatory variables. These can be set with certainty, or forecasted themselves, or you can work with assumptions (e.g., for scenario analysis). | |
| Jan 4, 2021 at 17:40 | comment | added | Numbermind | "If you fit a model using external variables and want to forecast from this model, you will need (forecasted) future values of the external variables, plain and simple. There is no way around this.". Can't you just forecast y if you know future values of the external variables? Imagine that external variables are weather variables. You can know those future values from weather forecast services. | |
| May 17, 2016 at 7:48 | history | answered | Stephan Kolassa | CC BY-SA 3.0 |