Evalvacija metod strojnega učenja pri napovedovanju finančnega propada podjetij

Šifrer, Ožbej

| | SLO | ENG | Cookies and privacy

Bigger font | Smaller font

First page > Show document

Show document

Title:	Evalvacija metod strojnega učenja pri napovedovanju finančnega propada podjetij
Authors:	ID Šifrer, Ožbej (Author) ID Kljajić Borštnar, Mirjana (Mentor) More about this mentor...
Files:	UN_Sifrer_Ozbej_2026.pdf (3,75 MB) MD5: FCA89B182F2F3715AEBC6F06C2F91BEB
Language:	Slovenian
Work type:	Bachelor thesis/paper
Typology:	2.11 - Undergraduate Thesis
Organization:	FOV - Faculty of Organizational Sciences in Kranj
Abstract:	Zaključno delo obravnava evalvacijo metod strojnega učenja pri napovedovanju finančnega propada podjetij ter raziskuje njihov potencial pri analizi finančnih podatkov. Namen dela ni razvoj produkcijskega napovednega modela, temveč poglobljeno razumevanje, kako različne metode strojnega učenja obravnavajo kompleksnost finančnih kazalnikov, kako stabilne so njihove napovedi in katere značilke imajo največji vpliv na prepoznavanje podjetij z večjim tveganjem propada. V raziskavi je preizkušen širši nabor metod strojnega učenja, ki se razlikujejo v pristopu k učenju, obravnavi nelinearnosti in občutljivosti na neuravnotežene podatke. Modeli so ovrednoteni z metrikami, kot so ROC AUC, natančnost in občutljivost, kar omogoča primerjavo njihove učinkovitosti in robustnosti. Poseben poudarek je namenjen interpretabilnosti modelov, ki je v finančnem sektorju ključnega pomena. Uporabljeni sta dve dopolnjujoči se metodi, in sicer SHAP analiza, ki omogoča vpogled v prispevek posameznih značilk k napovedi, ter ablacijska analiza, ki preverja odvisnost modelov od posameznih vhodnih spremenljivk. Delo poudarja, da strojno učenje predstavlja pomemben potencial pri analizi finančnega zdravja podjetij, saj omogoča odkrivanje skritih vzorcev, boljše razumevanje dejavnikov tveganja in razvoj interpretabilnih analitičnih orodij. Priporočila za nadaljnje delo vključujejo uporabo aktualnejših podatkov, razširitev nabora modelov, izboljšanje obravnave neuravnoteženosti in vključitev naprednih metod interpretabilnosti. Delo prispeva k boljšemu razumevanju možnosti uporabe sodobnih metod strojnega učenja pri podpori odločanja v financah in na področju upravljanja tveganj.
Keywords:	Strojno učenje, podatkovna analitika, finančni propad podjetij, interpretabilnost modelov, evalvacija metod
Place of publishing:	Maribor
Year of publishing:	2026
PID:	20.500.12556/DKUM-98795
COBISS.SI-ID:	286920963
Publication date in DKUM:	04.08.2026
Views:	228
Downloads:	22
Metadata:
Categories:	FOV
:	Copy citation

Average score:	(0 votes)
Your score:	Voting is allowed only for logged in users.
Share:

Hover the mouse pointer over a document title to show the abstract or click on the title to get all document metadata.

Licences

License:	CC BY 4.0, Creative Commons Attribution 4.0 International

Link:	http://creativecommons.org/licenses/by/4.0/
Description:	This is the standard Creative Commons license that gives others maximum freedom to do what they want with the work as long as they credit the author.
Licensing start date:	08.07.2026

Secondary language

Language:	English
Title:	Evaluation of machine learning methods in predicting corporate financial distress
Abstract:	This thesis examines the evaluation of machine learning methods for predicting corporate financial distress and explores their potential in the analysis of financial data. The aim of the study is not to develop a production‑ready predictive model, but rather to gain a deeper understanding of how different machine learning approaches handle the complexity of financial indicators, how stable their predictions are, and which features have the greatest influence on identifying companies with an increased risk of failure. The research evaluates a broad set of machine learning methods that differ in their learning strategies, ability to capture nonlinear relationships, and sensitivity to imbalanced data. The models are assessed using metrics such as ROC AUC, precision, and recall, enabling a comparison of their effectiveness and robustness. Particular emphasis is placed on model interpretability, which is crucial in the financial sector. Two complementary techniques are applied: SHAP analysis, which provides insight into the contribution of individual features to the prediction, and ablation analysis, which examines the dependence of models on specific input variables. The findings highlight that machine learning holds significant potential for analysing the financial health of companies, as it enables the discovery of hidden patterns, a better understanding of risk factors, and the development of interpretable analytical tools. Recommendations for future work include the use of more recent datasets, expanding the range of evaluated models, improving the handling of class imbalance, and incorporating advanced interpretability techniques. The study thus contributes to a deeper understanding of how modern machine learning methods can support decision‑making processes in finance and risk management.
Keywords:	machine learning, data analytics, corporate bankruptcy, model interpretability, method evaluation

Comments

Leave comment

You must log in to leave a comment.

Comments (0)

0 - 0 / 0

There are no comments!

Back