Metrics

The problem of choosing a metric for determining the degree similarities between objects of different nature (for strings, outliers, mixed and sparse data)

Sparse data

Initial class distribution and dictionaries used:

Research results

Identifying outliers in data

Source dataset:

Mahalanobis distance

The result of the application:

Cook's Distance

The result of the application:

DBSCAN method

The result of the application:

Mixed data

KNN

KMeans & K-Prototype

KMedoids