Data-Driven Market Analysis and Visualization for Supporting Strategic Decisions

Wajd Saeid, BSc

Diplomová práce

Abstract:

This thesis shows a data-driven analytical framework designed to support strategic decision-making by integrating data management, statistical analysis, visualization, and predictive modeling. The case study that used is video game market to explain how market data can be transformed to practical managerial insights. The main objective of the study was building a framework to help decision-makers to better understand market structure, identify sales patterns, evaluate influential factors, and asses potential market opportunities. The study’s practical part started by constructing relational database in MYSQL which organized the original dataset in a structured and normalized form. The database included tables for games, sales, genres, platforms, publishers, developers, and ratings. This layer of the database supported data consistency and provided the foundation for preprocessing and analytical tasks later. Afet the database stage, the data was cleaned, filtered, and prepared for analysis. The main analytical sample was taken is from 2000 to 2016 to focus on the most recent structure of the video game market. After that a descriptive analysis was made to summarize the information taken from the dataset and to provide the initial understanding of the market composition in terms of genre, platform groups, sales distribution, and review indicators. The inferential analytical stage used several statistical methods (t-test, ANOVA, chi-square test, and regression analysis) to examine whether differences and relationships in the data were statistically meaningful. And then several machine learning models were implemented and compared in the predictive stage like: Random Forest, Neural Network, and CatBoost. And that to choose the model with the best predictive performance for estimating sales. Finally, CatBoost was selected because it has strong suitability for the dataset, and also because of its ability to handle categorical variables. At the end a recommendation system was build based on the final CatBoost model. This system was designed as a hybrid decision-support tool that combines predicted sales with market indicators such as market demand, competition intensity, publisher position, and regional sales patterns. The predictive outputs were transformed into more practical strategic recommendations through this process, including identifying concepts, estimating risk, evaluating opportunity, and suggesting target regions. In addition, Microsoft Power BI was used to develop dashboards that present the main market in a visual way. These dashboards provided interactive views of sales trends, regional performance, genre distribution, platform evolution, publisher presence, and the relationship between reviews and sales. The dashboard layer complemented the analytical and predictive stage by making the results easier to interpret from managerial perspective. Overall, this thesis shows that combining database design, statistical analysis, machine learning, dashboard visualization, and a recommendation system can produce a practical analytical framework to support strategic decision-making. The findings suggest that such a framework can help managers to better understand historical market behavior and also evaluate hypothetical future opportunities in a more structured and evidence-based way.

Keywords

data-driven decision-making, market analysis, business intelligence, dashboard visualization

Jazyk práce: angličtina

Datum vytvoření / odevzdání či podání práce: 3. 4. 2026

Obhajoba závěrečné práce

• Obhajoba proběhla 1. 6. 2026
• Vedoucí: doc. Ing. Jan Tyrychtr, Ph.D.
• Oponent: Richard Vágner, externi

Česká zemědělská univerzita v Praze

Provozně ekonomická fakulta

Magisterský studijní program:
Economics and Management