I looked at various articles, and most of them talked about how to enter weight in a specific row of data, and not some feature. In R GBM, we can designate a model to study from one particular function more than another. Can we do something similar for Python XGBoost?
I went through the following:
here 1 here 2 here 3 and everyone somehow mentions that we can add weight to the row.
I am new to XGBoost and want a second confirmation if we can assign weight to variables or not.
I have a situation, I have planned columns and Target as Actuals. Whenever my Scheduled is below 5-10k, my model surpasses forecasts with a huge margin. It is disabled on average by 70% (average absolute error), while for planned> 10 thousand. My error metric is in the range of 20-30%. Is there a way to get XGBoost to train as planned?
I tried to introduce the bucket variable, the flag variable, but it turned out to be the least important function and therefore didn't help.
Or any other suggestion on how to apply this scenario would be really helpful.
If there is a way to assign weight (sample_weight) in XGBoost, then what the parameter (weight_data) should look like. His form and everything
thank