上海财经大学统计与管理学院邀请马铁丰教授作了一场题为“Double-K BS Algorithm for Multiple Change-Points Detection(双K BS算法的多变点检测)”,统计与管理学院主要的学科专业是统计学,包括经济管理统计、金融统计、统计理论与方法、数量金融与风险管理等多个学科方向。上海财经大学统计学科是一个历史悠久、成绩斐然的学科。讲座的主要内容是:
本文提出了一种改进的算法,命名为双K BS算法,它是基于BS算法。我们提出一个新的测试统计这是一个指数衰减加权累积和统计。结论通过与多个变化点检测的主要方法的比较,所提出的方法OERS的最佳性能。双K BS算法可以快速,准确地检测出多种变化点有时间复杂度低。此外,我们采用该方法在电力市场情况下降低其取得预期的效果。
原文:The paper propose an improved algorithm, named Double-K BS algorithm, which is based on BS algorithm. And we raise a new testing statistic which is an exponential decay weighted CUSUM statistic. Conclusion Through the comparison with the main methods for multiple change-points detection, the proposed method oers the optimal performance. Double-K BS algorithm can detect multiple change-points fast and accurately with a low time complexity. In addition, we apply the proposed method to scenarios reduction in electric market which obtains desired results.