Hi, i would like to build a custom objective funct...
# mlforecastj
jan rathfelder
06/19/2024, 1:17 PMHi, i would like to build a custom objective function for xgboost. but all stand solutions fail, because i cant really alter the fit method like i could do in xgboost (where i can just specify a custom objective function). i tried multiple ways but all failed so far. do you guys have any idea on how to do this? it seems one issue is that the train data that i can access is not in the Dmatrix style. i am happy for any suggestions here
jan rathfelder
06/19/2024, 1:45 PMwhen i do something simple like having this custom objective function:
_def_ mse(_self_, _preds_, _labels_):'''mse'''grad = 2.0 * (_preds_ - _labels_)hess = 2.0 * np.ones_like(_labels_)return grad, hessj
José Morales
06/19/2024, 2:58 PMHey. Have you tried this?
José Morales
06/19/2024, 3:06 PMOh nvm, I just read the optuna part
j
jan rathfelder
06/19/2024, 3:32 PMit is a bit weird, because when i work with dtrain like this, the error returned says they are numpy arrays not not Dmatrix objects, which i find starnge:
_def_ mse_jan(_self_, _preds_, _dtrain_):labels = _dtrain_.get_label()#black_friday_density = global_black_friday_density[dtrain.indices]#penalties = np.where(black_friday_density > 0, 1000, 1.0)grad = 2.0 * (_preds_ - labels) * penaltieshess = 2.0 * np.ones_like(labels)return grad, hessAn error occurred in validation step 1: 'numpy.ndarray' object has no attribute 'get_label'jan rathfelder
06/19/2024, 3:42 PMok, live update: one of these, probably th labels had to be converted to np. arrays. this seems to work now:
_def_ mse_jan(_self_, _preds_, _y_):#labels = dtrain.get_label()#black_friday_density = global_black_friday_density[dtrain.indices]#penalties = np.where(black_friday_density > 0, 1000, 1.0)grad = 2.0 * (np.array(_preds_) - np.array(_y_)) # * penaltieshess = 2.0 * np.ones_like(_y_)return grad, hess2 Views