I have heat consumption measurements in 6 facilities (monthly).
Based on data from 5 objects, I try to forecast consumption in one object. I need a forecast for the next 3 months. I am currently using random forest and got these results
What other approach do you propose? Can a more accurate forecast be expected?