ML developers by LLC Yandex

Machine Learning has garnered significant public interest in recent years, prompting professionals worldwide to acquire skills in this field.

Yandex Researches team undertook an educational infographic project in that field: they compiled the most-searched machine learning skills, and charted a graph with over 1,500 skills used by professionals in their work.

To achieve this, internet search statistics and the Stack Overflow database were used: search queries were matched with Stack Overflow tags, the results were grouped into clusters, and visualized.

The size of each skill on the map corresponds to its popularity among machine learning developers. The position of skills relative to each other is determined by the similarity of their context: the more frequently two skills are neighbour the same tags on Stack Overflow, the closer they are on the map. To assess contextual proximity, the vectors of mutual occurrence for each pair of skills were calculated and then normalized using the TF-IDF metric. The UMAP algorithm was used to organize the skills on the map.

The interactive map also provides additional information for each skill: a description; the dynamics of interest since 2010; the ten most frequent neighbors for each node; and a list of alternative skills that can replace the selected skill in certain tasks.

As a bonus, the most relevant skills of 2023 with particularly strong growth in search popularity over the year were highlighted.