Мультиклассовая классификация корпуса смешанных текстов алгоритмами машинного обучения

Abstract

This paper compares different machine learning algorithms for multiclass classification of mixed texts corpus. The F1-score was used as a quality metric for the algorithm comparison. The algorithms with acceptable quality for the corpus of texts were selected in the process. The dataset includes 7863 rows and 4 features, the gradient boosting showed the best result based on metric F1=0.771.