<p>Outlier detection is an important research topic in data mining. To address the difficulties in determining the minimum number of data objects and neighborhood radius, as well as the limited accuracy of density-based clustering outlier detection methods, this paper proposes a novel outlier detection algorithm based on the successive outlier factor. First, the natural neighbor search algorithm and a distance curve are employed to adaptively determine the parameter settings of the density-based clustering algorithm. Then, the DBSCAN algorithm is used to obtain candidate outliers. Subsequently, successive neighbors and successive span are defined to capture more accurate local information at smaller neighborhood scales while also reflecting global structural characteristics. Finally, a successive outlier factor is proposed to further quantify the degree of outlierness of data objects. Experimental results on both synthetic and real datasets demonstrate the effectiveness of the proposed algorithm in comparison with other outlier detection methods based on similar strategies.</p>