Michael Mayer
Michael Mayer
I like. Maybe worth noting that the one and only @david-cortes very recently removed non-dataset input in `xgb.cv()`.
But the default method works, right?
I get the same error. Any idea what's going wrong?
Great initiative, thanks a lot @trivialfis : 1. If @david-cortes has a little bit of spare time, he would be our man here. I can also help, but need a...
@hcho3: I wanted to propose {xgboost2} as well, but did not dare :-)
We would probably write the train/cv-API from scratch, but recycle the internals that wrap the C++ functions. I will start to study the current functions this weekend.
@david-cortes : I was a couple of days afk and will give my feedback soon. Thank you so much for starting this thread. We might get some inspiration of https://github.com/MathiasAmbuehl/modeltuner...
Curiously, I don't know the answers to the questions to me. But I think they are less important. My current mood: 1. Most of what have been suggested above means:...
@trivialfis Nice! So we need categorical support in R indeed at higher priority. In R, [shapviz](https://github.com/ModelOriented/shapviz) already works with categoricals, e.g., for LightGBM. So at least the plotting of SHAP...
TreeSHAP in XGBoost is calculated on "raw" scale, which is the link scale. Like Poisson and Gamma, Tweedie objective uses the log link, so the SHAP values will sum up...