Ecosyste.ms: Awesome

An open API service indexing awesome lists of open source software.

Awesome Lists | Featured Topics | Projects

https://github.com/bsharchilev/influence_boosting

Supporting code for the paper "Finding Influential Training Samples for Gradient Boosted Decision Trees"
https://github.com/bsharchilev/influence_boosting

catboost gradient-boosting influence-functions machine-learning machine-learning-algorithms paper python

Last synced: 2 months ago
JSON representation

Supporting code for the paper "Finding Influential Training Samples for Gradient Boosted Decision Trees"

Awesome Lists containing this project

README 

        
# Finding Influential Training Samples for Gradient Boosted Decision Trees

This repository implements the _LeafRefit_ and _LeafInfluence_ methods described in the paper [_Finding Influential Training Samples for Gradient Boosted Decision Trees_](https://arxiv.org/abs/1802.06640).



The paper deals with the problem of finding infuential training samples using the Infuence Functions framework from classical statistics recently revisited in the paper ["Understanding Black-box Predictions via Influence Functions"](https://arxiv.org/abs/1703.04730) ([code](https://github.com/kohpangwei/influence-release)). The classical approach, however, is only applicable to smooth parametric models. In our paper, we introduce _LeafRefit_ and _LeafInfuence_, methods for extending the Infuence Functions framework to non-parametric Gradient Boosted Decision Trees ensembles.



# Requirements

We recommend using the [Anaconda](https://www.anaconda.com/download/) Python distribution for easy installation. 

## Python packages

The following Python 2.7 packages are required:



_Note: versions of the packages specified below are the versions with which the experiments reported in the paper were tested._

- numpy==1.14.0

- scipy==0.19.1

- pandas==0.20.3

- scikit-learn==0.19.0

- matplotlib==2.0.2

- tensorflow==1.6.0rc0

- tqdm==4.19.5

- ipywidgets>=7.0.0 (for Jupyter Notebook rendering)



The ``create_influence_boosting_env.sh`` script creates the `influence_boosting` Conda environment with the required packages installed. You can run the script by running the following in the ``influence_boosting`` directory:

```shell

bash create_influence_boosting_env.sh

```



## CatBoost

The code in this repository uses [CatBoost](https://catboost.yandex/) for an implementation of GBDT. We tested our package with CatBoost version 0.6 built from [GitHub](https://github.com/catboost). Installation instructions are available in the [documentation](https://tech.yandex.com/catboost/doc/dg/concepts/python-installation-docpage/).



**_Note: if you are using the ``influence_boosting`` environment described above, make sure to install CatBoost specifically for this environment._**



## ``export_catboost``

Since CatBoost is written in C++, in order to use CatBoost models with our Python package, we also include ``export_catboost``, a binary that exports a saved CatBoost model to a human-readable JSON.



This repository assumes that a program named ``export_catboost`` is available in the shell. To ensure that, you can do the following:

- Select one of the two binaries, ``export_catboost_macosx`` or ``export_catboost_linux``, depending on your OS.

- Copy it to ``export_catboost`` in the root repository directory.

- Add the path to the root repository directory to the ``PATH`` environment variable.



**_Note: since CatBoost's treatment of categorical features can be fairly complicated, ``export_catboost`` currently supports numerical features only._**



# Example

An example experiment showing the API and a use-case of Influence Functions can be found in the [``influence_for_error_fixing.ipynb``](https://github.com/bsharchilev/influence_boosting/blob/master/scripts/influence_for_error_fixing.ipynb) notebook.



**_Note_**: in this notebook, CatBoost parameters are loaded from the [``catboost_params.json``](https://github.com/bsharchilev/influence_boosting/blob/master/data/adult/catboost_params.json) file. In particular, the ``task_type`` parameter is set to ``CPU`` by default. If you have a GPU with CUDA available on your machine and compiled CatBoost with GPU support, you can change this parameter to ``GPU`` in order to train CatBoost faster on GPU. The majority of the experiments in the paper were conducted using the ``GPU`` mode.