Risk control modeling framework(风控建模工具集)
声明:工具箱为风控建模过程提供支持,因工作紧迫暂于23Feb19停更;感谢jiangchao8@jd & huyao3@jd为这些积累提供若干支持
最新Note:
- 整个工具箱已做完re-view:包括每个模块其他语言的double-check、每个模块参数名称标准化
- 2.2.2 Chi2_merge.py模块可以使用,但还需修改
- Example.x为测试实例,可参照(!!实际项目中的特征名和数据等保密信息已做处理!!)
- Pypi:https://pypi.org/project/rcmf/
需更新:
- 特征筛选部分没有落地,仍在整理中,包括:集成模型筛选方法、逻辑回归L1正则化
- vintage、迁移率、pmml等
- 3个使用示例最好生成MD,增加可读性
- 2.2.2 Chi2_merge.py:去除卡方分箱的“最优分箱组数”判定,添加卡方值阈值约束
- 特征筛选:逐步回归
更新日志:
29Dec2018
- 添加特征筛选部分(2.x):包含特征稳定性(PSI from *Yao)、特征共线性(方差膨胀因子和相关性矩阵)
-
- Data_Explore.py:去除相关性矩阵、对分布图拼接按照4维或9维子图展示
- 4.1 Model performance.py:添加准召图
更新日志:
13Jan2019
- 2.2.2 Cut_merge.py:① 优化分箱过程,直接从pd.cut解析结果; ② 优化从分箱数据集至WOE数据集的映射逻辑; ③ 将WOE计算并入分箱程序
- 4.1 Model performance.py:① 利用sklearn中混淆矩阵函数加快计算; ② 添加3线图(准确率、召回率、F1-Score)
更新日志:
25Jan2019
- 2.x Collinearity.py:相关性计算部分的热力图展示,可以控制哪些特征用于绘图;例如可以abs(corr) > 0.8的特征,或者是按相关性排序后取前20个特征
- 3.x PDO_Score_Convert.py:放活odds定义,如不定义,则按照如下规则计算:对train数据集计算ks切点,取切点的百分位点区间内的所有y_label,计算好坏比
- RCMF使用说明:完成初版
更新日志:
10Jun2019
- 已为全部脚本更新help
-
- Data_Explore.py:更新计算方式,更新数值型特征的自定义分布,更新伪字符型变量(99%都是数值的字符型变量)分布探索
Metadata
Release files for rcmf 1.0.6
For a detailed explanation of source distributions (sdists) and built distributions (wheels), please see the package formats documentation.
Source distribution (sdist)
| File | Size | Uploaded | Provenance | Actions |
|---|---|---|---|---|
| rcmf-1.0.6.tar.gz | 26.9 kB | Without provenance | Details for rcmf-1.0.6.tar.gz |
Built distribution (wheel)
| File | Interpreter | ABI | Platform | Actions Reset |
|---|---|---|---|---|
| rcmf-1.0.6-py3-none-any.whl | Python 3 | none | any | Details for rcmf-1.0.6-py3-none-any.whl |
Total release size: 57.6 kB
Release files / rcmf-1.0.6.tar.gz
| Download URL | rcmf-1.0.6.tar.gz |
|---|---|
| Size | 26.9 kB |
| Tags | Source |
|
SHA-256 checksum How to use checksums |
694db916aad9cda120a8efdc0476bc9ec22cfb8e5e4a3e8a6998507156e24a1c
|
|
BLAKE2b-256 checksum How to use checksums |
d18f305da0f3fe680ab545bcb676ef431e99a9393b6b9bda3d9853cd7fcca318
|
| Upload date | |
|
Uploaded using Trusted Publishing? What is trusted publishing? |
No |
| Uploaded via |
twine/1.13.0 pkginfo/1.4.2 requests/2.18.4 setuptools/40.8.0 requests-toolbelt/0.9.1 tqdm/4.28.1 CPython/3.6.5
|
Release files / rcmf-1.0.6-py3-none-any.whl
| Download URL | rcmf-1.0.6-py3-none-any.whl |
|---|---|
| Size | 30.7 kB |
| Tags | Python 3 |
|
SHA-256 checksum How to use checksums |
e3237dda140a9dc94d7dc8aa98bbf448e8561dc5fa9058c5bf79eac43dedff5e
|
|
BLAKE2b-256 checksum How to use checksums |
787013428cf5817c417d77f1971386db45a12c593d0b16ff1e3e344d163c32d7
|
| Upload date | |
|
Uploaded using Trusted Publishing? What is trusted publishing? |
No |
| Uploaded via |
twine/1.13.0 pkginfo/1.4.2 requests/2.18.4 setuptools/40.8.0 requests-toolbelt/0.9.1 tqdm/4.28.1 CPython/3.6.5
|
Security for rcmf 1.0.6
Provenance
Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.
No provenance found
None of the files in this release were uploaded with provenance metadata.
A note on package security
PyPI is a repository for distributing packages, not a service for auditing them. While third-party security researchers and our staff do their best to detect and remove malicious packages, PyPI cannot guarantee the quality, security or suitability of any given package.
To learn more about protecting yourself from malicious actors, visit our package security documentation.