基于趋势秩的Spearman相关方法

Spearman Rank Correlation Method Based on Trend Rank

针对Spearman秩相关方法在两个变量局部数据的大小次序不一致时描述变量间趋势相关性效果不佳的问题,提出了基于趋势秩的Spearman相关方法(T-SRC).T-SRC设计了将数据转换为趋势秩的方法,专门捕获数据的变化趋势,从而提高了Spearman相关方法发现变量间趋势相关的性能.真实数据的实验结果表明,与传统的Spearman秩相关方法相比,T-SRC挖掘变量间的趋势相关关系的性能更优,验证了方法的有效性.

The Spearman rank correlation method has poor performance when it is used to find trend correlations between variables, where the local inconsistent ranks appear. The trend-rank Spearman correlation method （T-SRC） is proposed to solve this problem. The method of transforming data to trend ranks is designed especially to catch the changing tendency of data, so that this can enhance the performance of finding trend correlations between variables. The experimental results on real data sets show that the T-SRC has better performance on mining trend correlations between variables, compared with traditional Spearman rank correlation method.