LSC-Mine: Algorithm for Mining Local Outliers

Data objects which differ significantly from the remaining data objects are referred to as outliers. Density-based algorithms for mining outliers are very effective in detecting all forms of outliers, where data objects with fewer neighbors are likely to be outliers than are those with more neighbors. However, existing density-based algorithms engage in huge repetitive computation and comparison for every object before the few outliers are detected. Expensive computations might make scalability of these techniques to important applications like quick fraud detection unfeasible. This paper proposes LSC-Mine algorithm based on the distance of an object and those of its knearest neighbors. In addition, data objects that are not possible outlier candidates are pruned which reduces the number of computations and comparisons in LOF technique resulting in an improved performance.