Multi-Variate Time-Series forecasting with XGBoost

Question

I have trained an XGBoost model on a time-series dataset for predicting a value. The time series has 5 features and one label (the target value). The trained model works fine on both training and testing data, so far so good. As I said, this dataset has some features that I have used for training the XGBoost model (i.e. a multi-variate dataset). The problem is that currently, I have values of these 5 features in my current dataset, so I can train the model with, and do the testing as well. But, I do not know these features values in future.

My question is, how can I predict the target value for future (Ex. next year) When I don't know the values of features in future to feed them into the trained model to do the prediction.

Answer 1

Simple answer- you can't.

If you can predict the values of these features, then you can see what the output of the model will be given those possible features. You can try and train a model on the t+1 of a feature, but whether this would work really depends on the nature of your data.

More nuanced answer:

Circumstances where you might be able to predict them include where there is periodicity, trend, or some other underlying structure, that makes the future look like the past. In this case, linear-style methods such as vector autoregression or dynamic mode decomposition (particularly where there are oscillating modes) might be better tools for capturing these behaviours explicitly.