IMBALANCE OF CLASSES IN SOLVING THE PROBLEM OF SOCIAL NETWORKS USER CLASSIFICATION FOR PROFESSIONAL ORIENTATION

Abstract

The problem of data imbalance is often underestimated when solving classification problems. A classification model that looks well trained on your data and gives a good recognition rate may not be reliable. Consideration of this problem in the specific task of classifying users of social networks will make it possible to understand how, why and, most importantly, when it is necessary to get rid from data imbalances.