-
## Description
I have a trained model with a binary objective using n_estimators=1000. The model performance (AUC) is quite good. I need the raw probabilities for selection by ranking. The probabilit…
-
During our work, we have encountered misunderstandings with the serializers type that we used from your schema (interface.json). It works fine when we work with "dataframe" type, but when we try to ob…
-
CI test **linux://doc/source/train/examples/lightgbm:lightgbm_example** is consistently_failing. Recent failures:
- https://buildkite.com/ray-project/postmerge/builds/6616#0192b1bf-9b58-4a3a-9a4d-29…
-
Improve your readme and include data description, links (article, contact, etc), etc;
Add interpretation/observation to your plots;
Use sklearn encoder to encode the features;
Please remove the use…
-
It would be nice if there was a way to add custom loss functions to ranger (e.g. poisson) for model training.
-
## Description
Changing leaf values with `set_leaf_output` in callback doesn't work the same as using `lgb.train` for each epoch.
## Reproducible example
```python
X, y = make_regressio…
-
I'm using SHAP with LightGBM for regression.
The problem is that SHAP values don't sum up the predicted value.
Predicted values are all in the 0 - 10000 range. I have ~9000 features.
![Screen…
-
## 🐛 Bug Description
## To Reproduce
Steps to reproduce the behavior:
1. take any yaml file under examples/benchmark. I'll take workflow_config_lightgbm_Alpha158.yaml for example.
2. Ch…
-
continuing from #79
In splitting.py, the left/right_indices_buffer will use up 8GB for 10^9 rows. If that causes swapping, the performance benefit of multithreading (which requires these buffers) a…
-
Many models (LightGBM etc) have recently been converted to explicitly running with n_jobs = physical cores not virtual cores, as recommended by those model maintainers for best performance.
`Random…