2025.5.25

1. Disclaimer: If you intend to modify this code, you must possess a solid mathematical foundation sufficient to derive the SVM process, as it involves the interpretation of the kernel.
2. Below is the proof of SVM (if you wish to delve deeper, you may refer to the LIBSVM toolbox by Professor Chih-Jen Lin's team, which includes demos for experimentation):
   
   
   
   
3. The detailed derivation process and experimental data can be found in the following paper: The Utility of Hyperplane Angle Metric in Detecting Financial Concept Drift.
4. Special Note: Our objective is to identify the degree of variation in classification boundaries within the space, not to focus on financial tasks or the properties of SVM.
5. Our method can be extended to any machine learning approach.
6. The LIBSVM toolbox can be installed and run in MATLAB.
7. Some researchers have already executed my code and provided positive feedback, including a Ph.D. candidate from Wuhan University and a quantitative researcher in the U.S. who graduated from Peking University.
8. If you encounter errors, have questions, or wish to share insights while running the code, email [email protected].
9. DOWNLOAD DATA: https://zenodo.org/records/15515662?token=eyJhbGciOiJIUzUxMiJ9.eyJpZCI6IjRjOThmMzRmLTJiMzYtNDFiOC1hMzAxLTI5NGE2YTNjM2M3ZCIsImRhdGEiOnt9LCJyYW5kb20iOiIxYTJjNTBiYzAzMmJmMGFlNjU2NTlkZjdhMmZlOTVkMiJ9.8XZaxBtcoxLDJcgSdNK20aQwoF1T0VElDpaOxg1tVioUzjk1CPP_SNmIhLnj0aiagpZ4qTu5EMPHaj_MZ66t4w