By Zheming Zuo^1,*†, Jie Li^2,*, Han Xu^3,4, and Noura Al Moubayed^1
1 Department of Computer Science, Durham University, Durham DH1 3LE, UK
² School of Computing, Engineering & Digital Technologies, Teesside University, Middlesbrough TS3 6DR, UK
³ Institute of Microelectronics, Chinese Academy of Sciences, Beijing 100029, China
⁴ University of Chinese Academy of Sciences, Beijing 100049, China
^* Equal contribution.
^† Corresponding author.

This is an official implementation of our Curvature-based Feature Selection (CFS) method.

CFS is a simple yet efficient feature selection method, which perfroms based on the Menger curvature and contributes the classification performance.

For more details, please refer our paper.

1. Requirement
2. Preparation
3. Run
4. Performance

• MATLAB >= 2016a

Clone the github repository. We will call the directory $CFS_ROOT

  git https://github.com/zhemingzuo/CFS
  cd $CFS_ROOT

Run our CFS method

  cd $CFS_ROOT/src

and then run demo_CFS.m.

For performance comparisons that might be used in your ongoing research work, we list the peak performance of CFS on four healthcare datasets:

• Cervical Cancer (Risk Factors) Data Set (CCRFDS)
• Breast Cancer Coimbra Data Set (BCCDS)
• Breast Tissue Data Set (BTDS)
• Diabetic Retinopathy Debrecen Data Set (DRDDS).

... denotes that our CFS can be combined with multiple supervised classifiers to achieve the same classification performance.

If you find CFS useful in your research, please consider citing:

ArXiv version:

@article{zuo2021cfs,
	title = {Curvature-based Feature Selection with Application in Classifying Electronic Health Records},
	author = {Z. Zuo and J. Li and H. Xu and N. A. Moubayed},
	journal = {arXiv preprint arXiv:2101.03581},
	year = {2021}
}

Journal version:

@article{zuo2021cfs,
	title = {Curvature-based Feature Selection with Application in Classifying Electronic Health Records},
	author = {Z. Zuo and J. Li and H. Xu and N. A. Moubayed},
	journal = {Technological Forecasting and Social Change},
	volume = {173},
	pages = {121--127},
	issn = {0040-1625},
	doi = {https://doi.org/10.1016/j.techfore.2021.121127},
	year = {2021}
}