Rolling Time Series Predictions - forecasting unique IDs using Gradient Boost

Hi there!

Thanks to @alexanderfillbrunn Rolling Time Series Predictions workflow, I was able to predict new values based on past values.I have also slightly adjusted the workflow to use Gradient Boost predicition (regression) instead of polynomial.

However,

My current use case : finding out future Mbps values based on past values for different cell IDs.

Problem: no matter how long I put the lag, I’m still getting the predicition for the last cellID i have in my input excel sheet.

Attached is the a snapshot of output i’m recieving plus my workflow.