Insight on 'date(actual)' Role in Time-aware Models
Hello DataRobot Community,
I am currently using DataRobot's time-aware modeling capabilities for regression analysis and forecasting. I've noticed a consistent pattern where the automatically generated "date(actual)" feature frequently emerges as a highly influential factor in the models' predictive power.
While I understand the general concept of time-aware modeling, I am seeking a deeper understanding of why and how the "date(actual)" feature impacts the predictive capability of the models in this specific context.
Can anyone point me to detailed documentation or other resources that explain:
The mechanisms underlying the high influence of the "date(actual)" feature in these time-aware models?
How DataRobot's algorithms determine the importance of "date(actual)" in these models?