I have built a forecasting model based on environmental variables. I didn't include economical variables. It worked fine. But in a new Covid reality model is overestimating all predictions due to economic impact. The training was conducted over a 5-year period data. Is there any option to rebuild NN to put more emphasis on last year's data? Probably feeding a model with data from 2020 several times make sens, but how can I get better results without manipulating with data?
1.4m articles
1.4m replys
5 comments
57.0k users