Bachelor's Thesis

Application of Mathematical and Statistical Methods in Company Management Using a Data-Driven Real Estate Platform

Final Thesis 1.18 MB Appendix 299.04 kB

Author of thesis: Jakub Kufčák

Acad. year: 2025/2026

Supervisor: Ing. Karel Doubravský, Ph.D.

Reviewer: Ing. Jakub Ulč

Abstract:

This bachelor's thesis applies selected mathematical and statistical methods to support managerial decision-making in a small Czech company operating the data-driven real estate platform realitnistrazce.cz. The work designs and empirically verifies a hedonic price model for residential apartments offered for sale in Czechia. The methodology covers data collection from the platform database, data cleaning, exploratory analysis, multiple linear regression in a log-linear specification, model validation through five-fold cross-validation and residual diagnostics, and the identification of underpriced and overpriced listings based on model residuals. The model is estimated on a sample of 143,510 cleaned apartment listings and explains roughly 71% variance in log asking price, with regional location, building condition and floor area emerging as the strongest price drivers, in that order. The thesis translates these statistical results into concrete managerial recommendations covering pricing analytics, lead generation and data governance for the platform.

Keywords:

hedonic price model, multiple linear regression, real estate, data quality, exploratory data analysis, residual diagnostics, cross-validation, managerial decision-making, key performance indicators, czech residential market, automated valuation

Date of defence

18.06.2026

Result of the defence

Defended (thesis was successfully defended)

znamkaBznamka

Grading

B

Process of defence

In his presentation, the student informed the committee about the objectives, solutions and results he had reached in his thesis. The committee then read the opinions and evaluation of the thesis supervisor and the opponent. The student answered the questions from the supervisor's assessment in full, the questions from the opponent's assessment in full. Questions from committee members: 1. Ing. Ulč: When you are going to launch the app? - answered 2. Ing. Ulč: What are the external risks? - answered On the basis of the presentation and the answers to the questions asked in the discussion, the committee decided that the student defended the thesis.

Language of thesis

English

Faculty

Fakulta podnikatelská

Department

Institute of Management

Study programme

Entrepreneurship and Small Business Development (BAK-ESBD)

Composition of Committee

doc. Ing. Robert Zich, Ph.D. (předseda)
doc. Ing. Pavla Marciánová, Ph.D. (místopředseda)
Ing. David Havíř, Ph.D. (člen)
Ing. Jakub Ulč (člen)
Ing. David Schüller, Ph.D. (člen)

Supervisor’s report
Ing. Karel Doubravský, Ph.D.

This bachelor’s thesis is a high-quality analytical project that successfully combines advanced statistical modeling with the real-world needs of business management. The author has developed a robust methodological framework for real estate valuation. Specific proposals are elaborated across three key areas. The main contribution lies in the transformation of raw data into “actionable management indicators”. Based on the above, I recommend this thesis for defense.

Questions:

1. How do you plan to incorporate the impact of specific local amenities (such as proximity to the subway or parks) into the model in the future in order to further improve its predictive power?

2. What are the main risks of relying on an automated hedonic model for customer acquisition in regions where there is insufficient historical transaction data?
Evaluation criteria Verbal classification Grade
Splnění stanovených cílů The objective of this thesis, which focused on the design and practical validation of mathematical and statistical methods to support the management of the real estate platform realitnistrazce.cz, was fully achieved. The author successfully developed and validated a functional hedonic pricing model, which he translated into specific management recommendations. A
Zvolený postup řešení, adekvátnost použitých metod The chosen methodology, which employs multiple linear regression in a log-linear specification and advanced techniques such as five-fold cross-validation, is highly appropriate for this purpose. I particularly appreciate the emphasis on data cleaning and residual diagnostics. A
Schopnost interpretovat dosažené výsledky a vyvozovat z nich závěry The student demonstrated excellent analytical skills in identifying key price determinants (location, building condition, floor area) and interpreting the variance in the data. The conclusions drawn are logically supported by statistical tests, and the author does not shy away from discussing the model’s limitations. A
Praktická využitelnost výsledků This work has exceptional practical value, as the proposed Python-based workflow can be directly integrated into the platform’s operational stack. Identifying undervalued and overvalued listings using model residuals provides the platform with a direct tool for customer acquisition and pricing analytics. A
Uspořádání práce, formální náležitosti, použitá terminologie a odborná jazyková úroveň The thesis is logically structured and well-organized; its layout and rigorous approach make it resemble a scientific article more than a typical thesis. Given the quality of the work, it is a pity that the hyphen, dash, and minus signs are not always used correctly. Also, the captions for some of the graphs are in Czech. B
Práce s informačními zdroji, včetně citací The author draws on current and relevant academic literature as well as online sources, all of which are properly cited in the text.  B

Grade proposed by supervisor: A

Reviewer’s report
Ing. Jakub Ulč

The aim of the thesis was to design and verify the use of mathematical and statistical methods to support company management using a data-driven real estate platform. The aim is considered fulfilled. The theoretical part adequately covers hedonic price models and regression analysis. The analytical part demonstrates solid statistical work — the model is estimated on an extensive dataset of cleaned listings with proper cross-validation and residual diagnostics. However, the author discloses being the sole founder of realitnistrazce.cz, and the platform's data infrastructure was operational before the thesis commenced (in my opinion). It remains unclear which specific analytical components were developed during the thesis period versus what existed as part of the platform's prior operations. While the statistical methodology is sound, the contribution to practical applicability is limited given the pre-existing nature of the business. The managerial recommendations lack specificity regarding implementation. I recommend the thesis for defense.
Evaluation criteria Grade
Splnění stanovených cílů B
Zvolený postup řešení, adekvátnost použitých metod B
Schopnost interpretovat dosažené výsledky a vyvozovat z nich závěry B
Praktická využitelnost výsledků C
Struktura práce, použitá terminologie a odborná jazyková úroveň B
Práce s informačními zdroji B
Topics for thesis defence:
  1. Which specific analytical features presented in this thesis were developed during the thesis period, and which were already implemented in the platform before you began writing?

Grade proposed by reviewer: B

Responsibility: Mgr. et Mgr. Hana Odstrčilová