Outlier Detection Model In Outliers

There are various models used in outliers. The interpretability of an outlier detection model is extremely important from the perspective of the analyst [2]. It is often desired to regulate a specific point as an anomaly in terms of its performance with other data. It provides the hints to the analyst about the diagnosis required in an application specific scenario and this is also referred as the intentional knowledge about the outliers [11]. Figure 4 displays the different outlier detection models.

Figure.4.Outlier detection models
4.1 Extreme value analysis Extreme value analysis is one of the basic forms of outlier detection for one dimensional data. This type of outlier is simulated as when the value is too high or too low.…show more content…
Dimensionally all regions become sparse in density model. The sparsity behavior in high dimensionality makes all points look very similar to one another and the observation is true and these outliers may be discovered by examining the distribution of the data in a lower dimensional local subspace [2].Low dimensional subspace contains outlier and the character is latent to analysis. This is the combinational process of clustering and regression. To show the outlier pattern analysis and subspace analysis high dimensional model is determined. This can be a huge challenge to compute the high dimensional data because the simultaneous discovery of relevant data localities and subspaces is very difficult [2].High-dimensional methods provide an interesting direction for intentional understanding of outlier analysis when the subspaces are described in terms of the original attributes…show more content…
It works efficiently in larger databases. It takes only the sample datum from the dataset instead of full dataset [17].It selects the data from the set and finds medoids for finding the outlier. The dataset D is taken as input and the sample data is taken as S where the PAM concept is applied to CLARA to find the medoid M. After finding the medoid the entire dataset D is classified based on cost then the average is calculated to find the dissimilarities between the clusters.

CLARA Pseudo code

5.3 LOCAL OUTLIER FACTOR This algorithm is proposed by Markus M. Breunig, Hans-Peter Kriegel [18], Raymond T. Ng and Jorg Sander in 2000 for finding anomalous data points by measuring the local deviation of a given data point with respect to its neighbors [18]. Higher amounts of data are collected and they are stored in databases by increasing the need of the data for efficient and effective analysis methods to make use of the information contained implicitly in the data [18].LOF is found to the every data in the dataset to indicate the degree of outlierness. The outlier factor is local in the sense that only a restricted neighborhood of each object is taken into an account [18].The reachability distance should be found from the centroid to define the