# Home Credit Default Risk Competition
This repository has a complete pipeline for predicting default risk on [Home Credit Default Risk Competition](https://www.kaggle.com/c/home-credit-default-risk) at Kaggle.
This model is part of the [7th place solution](https://www.kaggle.com/c/home-credit-default-risk/discussion/64580), which consists of an ensemble of 13 different models using linear regression.
In this Kaggle competition, participants must predict if a customer will default on a given credit application (binary classification).
The effectiveness of a solution was evaluated with the AUC ROC metric, which is very common in this kind of problem.
This single model is able to achieve the 32nd position on the [leaderboard](https://www.kaggle.com/c/home-credit-default-risk/leaderboard),
with a 0.8044 private score.
The LightGBM library is used to implement a gradient boosting decision tree algorithm with Gradient-based One-Side Sampling (GOSS).
This novel technique was introduced by the Microsoft Research team and is described in [this paper](https://papers.nips.cc/paper/6907-lightgbm-a-highly-efficient-gradient-boosting-decision-tree.pdf).
## Installation
1. Install Python 3 with pip
2. Clone repository and install requirements
```pip3 install -r requirements.txt```
3. Change the path variables at the config.py file
4. Run the main file
```python main.py```
### Dependencies
- Python 3
- Numpy
- Sklearn
- Lightgbm
- Hyperopt
The code was tested with Python 3.6 on a Windows 7 machine and Python 3.5 on Ubuntu 16 LTS.
Kaggle竞赛中预测信用违约风险的梯度提升模型.zip
版权申诉
156 浏览量
2023-03-31
23:12:03
上传
评论 2
收藏 24KB ZIP 举报
快撑死的鱼
- 粉丝: 1w+
- 资源: 9156
最新资源
- 历届(第1-21届)希望杯数学竞赛初一试题及答案(最新整理).doc全国数学邀请赛(264页资料)
- 水滴.psd
- TokenPocket_V2.1.2_release.apk
- Apache-druid-kafka-rce.yaml
- 半导体行业词汇 相关缩写
- 基于C#的ASP.NET数据库原理及应用技术课程指导平台的开发
- 基于ROS的智能车轨迹跟踪算法的仿真与设计源码运用PID跟踪算法.zip.zip
- Bug Bounty Tip - i春秋Self-XSS变废为宝的奇思妙想
- 1991-2015年全国初中化学竞赛复赛试题汇编(212页)(24年竞赛复赛真题).docx天原杯
- Apache Flink 未授权访问+远程代码执行.pdf
资源上传下载、课程学习等过程中有任何疑问或建议,欢迎提出宝贵意见哦~我们会及时处理!
点击此处反馈