[Retracted] Effect of Improved Association Algorithm on Mining and Recognition of Audit Data

Algorithm 1

Cell-based outlier detection algorithm flow.

(1)	The algorithm inputs all data objects, and calculates its distance threshold r according to formula (8).
(2)	The algorithm divides the cells according to formula (9), records the number of objects in each cell as count, and assigns the initial value count = 0.
(3)	The algorithm assigns each object to the corresponding cell according to formula (10), and makes the count++ of the corresponding cell.
(4)	The algorithm determines the threshold of the number of points in the neighborhood according to formula (11).
(5)	The algorithm repeats steps (6)–(18) for each cell .
(6)	The algorithm calculates the dynamic threshold according to formula (4).
(7)	If corresponds to , then all objects in are not isolated points, and the algorithm goes to step (5).
(8)	, then all objects in are not outliers, and the algorithm goes to step (5), where represents the number of first-level units of .
(9)	, then all objects in are isolated points, mark all objects in as Red, and the algorithm goes to step (5), where represents the number of the second layer unit of .
(10)	Else: The algorithm repeats steps (11) –(17) for each object P in .
(11)	We assume that the number of objects in the r neighborhood of object P is , and initialize it by .
(12)	The algorithm repeats steps (13) –(15) for each object Q of the second level unit of .
(13)	The algorithm calculates the distance between P and Q.
(14)	, then .
(15)	The algorithm goes to step (12).
(16)	, then the object P is an isolated point, which is marked as Red.
(17)	The algorithm goes to step (10).
(18)	The algorithm goes to step (5).
(19)	The algorithm removes all objects marked as Red as all outliers.