Intelligent System for Employee Performance Prediction in Large IT Companies
Students Name: Vorozhko Vitalii Anatoliiovych
Qualification Level: magister
Speciality: Information Control Systems and Technologies
Institute: Institute of Computer Science and Information Technologies
Mode of Study: full
Academic Year: 2025-2026 н.р.
Language of Defence: ukrainian
Abstract: Relevance. Traditional methods for assessing the productivity of employees of IT companies are characterized by subjectivity, inefficiency and data dispersion, which makes it impossible to make timely management decisions and identify risks of reducing personnel efficiency. The object of the study is the process of assessing and predicting the productivity of employees of large IT companies. The subject of the study is machine learning methods for automated forecasting of personnel productivity based on work activity metrics and HR data. The purpose of the study is to develop an intelligent system for predicting the productivity of employees of IT companies using machine learning algorithms to increase the objectivity and efficiency of personnel assessment. Structure of the work. This master’s work consists of an introduction, four sections, conclusions, a list of sources used and appendices. The first section substantiates the relevance of the topic, analyzes the subject area, modern approaches to performance forecasting. The second section formulates the problem, selects machine learning methods and technologies. The third section describes the development technologies and tools, implements the ML-pipeline, the database structure and the web interface of the system. The fourth section presents the results of experimental research, which presents the results of a comparative analysis of algorithms, hyperparameter optimization, feature importance analysis and model validation on real data. Research methods and tools. The work uses machine learning algorithms (XGBoost, Random Forest, Gradient Boosting, Logistic Regression), data preprocessing methods (StandardScaler, OneHotEncoder), hyperparameter optimization via Grid Search, cross-validation, feature importance analysis, Python, Flask, MySQL, scikit-learn technologies, HTML/CSS/JavaScript/Bootstrap web technologies. Results and practical significance. An intelligent performance forecasting system with a classification accuracy of 89.4% and a determination coefficient of R?=0.838 has been developed, which allows automating personnel evaluation, reducing the time for preparing analytical reporting by 3-5 times, and providing objective support for management decisions in large IT companies. The full scope of the work is 108 pages, including 71 pages of the main text, contains 19 figures, 5 tables, and the list of sources used includes 49 items. Keywords: intelligent system, machine learning, performance forecasting, data analytics, Python.