https://github.com/artefactory/mgs-grf
https://github.com/artefactory/mgs-grf
research-center
Last synced: about 1 year ago
JSON representation
- Host: GitHub
- URL: https://github.com/artefactory/mgs-grf
- Owner: artefactory
- License: mit
- Created: 2024-07-09T14:27:57.000Z (about 2 years ago)
- Default Branch: main
- Last Pushed: 2025-05-13T12:53:10.000Z (about 1 year ago)
- Last Synced: 2025-05-13T13:43:48.038Z (about 1 year ago)
- Topics: research-center
- Language: Python
- Homepage:
- Size: 3.34 MB
- Stars: 6
- Watchers: 0
- Forks: 0
- Open Issues: 0
-
Metadata Files:
- Readme: README.md
- License: LICENSE
Awesome Lists containing this project
README
# Harnessing Mixed Features for Imbalance Data Oversampling: Application to Bank Customers Scoring
Abdoulaye SAKHO1, 2, Emmanuel MALHERBE1, Carl-Erik GAUTHIER3, Erwan SCORNET2
1 Artefact Research Center,
2 *LPSM* - Sorbonne Université, 3 Société Générale
Preprint.
[[Full Paper]](https://arxiv.org/pdf/2503.22730)
**Abstract:** *This study investigates rare event detection on tabular data within binary classification. Many real-world classification tasks, such as in banking sector, deal with mixed features, which have a significant impact on predictive performances. To this purpose, we introduce MGS-GRF, an oversampling strategy designed for mixed features. This method uses a kernel density estimator with locally estimated full-rank covariances to generate continuous features, while categorical ones are drawn from the original samples through a generalized random forest.*
You will find code to reproduce the paper experiments as well as an nice implementation of our *new* and *efficient* strategy for your projects.
## ⭐ Table of Contents
- [How to use MGS-GRF](#-how-to-use-the-mgs-grf-algorithm-to-learn-on-imbalanced-data)
- [Reproducing the paper experiments](#-reproducing-the-paper-experiments)
- [Data sets](#-data-sets)
- [Acknowledgements](#-acknowledgements)
- [Citation](#-citation)
## ⭐ How to use the MGS-GRF Algorithm to learn on imbalanced data
Here is a short example on how to use MGS-GRF:
```python
from mgs_grf import MGSGRFOverSampler
## Apply MGS-GRF procedure to oversample the data
mgs_grf = MGSGRFOverSampler(K=len(numeric_features),categorical_features=categorical_features,random_state=0)
balanced_X, balanced_y = mgs_grf.fit_resample(X_train,y_train)
print("Augmented data : ", Counter(balanced_y))
## Encode the categorical variables
enc = OneHotEncoder(handle_unknown='ignore',sparse_output=False)
balanced_X_encoded = enc.fit_transform(balanced_X[:,categorical_features])
balanced_X_final = np.hstack((balanced_X[:,numeric_features],balanced_X_encoded))
# Fit the final classifier on the augmented data
clf_mgs = lgb.LGBMClassifier(n_estimators=100,verbosity=-1, random_state=0)
clf_mgs.fit(balanced_X_final, balanced_y)
```
A more detailed notebook example is available [here](example/example.ipynb).
## ⭐ Reproducing the paper experiments
If you want to reproduce our paper experiments:
- Section 4.2 : the [Python file](protocols/run_synthetic_coherence.py) reproduce the experiments (data sets, oversampling and traing). Then the results can be analyzed with this [notebook](protocols/notebooks/res_coh.ipynb).
- Section 4.3 : the [Python file](protocols/run_synthetic_association.py) reproduce the experiments (data sets, oversampling and traing). Then the results can be analyzed with this [notebook](protocols/notebooks/res_asso.ipynb).
- Section 5 : the [Python file](protocols/run_protocol-final.py) reproduce the experiments (data sets, oversampling and traing). Then the results can be analyzed with this [notebook](protocols/notebooks/res_real_data.ipynb).
## ⭐ Data sets
The data sets of used for our article should be dowloaded inside the *data/externals* folder. The data sets are available at the followings adresses :
* [BankMarketing](https://archive.ics.uci.edu/dataset/222/bank+marketing)
* [BankChurners](https://www.kaggle.com/datasets/thedevastator/predicting-credit-card-customer-attrition-with-m)
## ⭐ Acknowledgements
This work was done through a partenership between **Artefact Research Center** and the **Laboratoire de Probabilités Statistiques et Modélisation** (LPSM) of Sorbonne University.
## ⭐ Citation
If you find the code usefull, please consider citing us :
```
@article{sakho2025harnessing,
title={Harnessing Mixed Features for Imbalance Data Oversampling: Application to Bank Customers Scoring},
author={Sakho, Abdoulaye and Malherbe, Emmanuel and Gauthier, Carl-Erik and Scornet, Erwan},
journal={arXiv preprint arXiv:2503.22730},
year={2025}
}
```