03325ntm a22005057i 4500 000725042 CZ-PrVSE 20251011201509.0 m d cr n|||||||||| 251011s2025 xr fsbm 000 0 eng d NEZPRACOVANÝ IMPORT ABA006 cze ABA006 ABA006 rda Lušková, Martina ISIS:170748 dis Boosting prediktivního modelu pomocí exception rules eng Boosting Predictive Models with Exception Rules / Martina Lušková 2025 ?? stran : digital, PDF soubor Vedoucí práce: Petr Máša Diplomová práce (Ing.)—Vysoká škola ekonomická v Praze. Fakulta informatiky a statistiky, 2025 Obsahuje bibliografii Textový (vysokoškolská kvalifikační práce) Rok obhajoby 2025 This thesis presents a novel hybrid approach to predictive modelling that enhances logistic regression with exception rules. While regression-based models are widely used for their simplicity and generalisation ability, they often fail to capture localised, non-linear patterns. To address this limitation, the proposed algorithm integrates rule-based refinement into a standard logistic regression model by mining exception rules using the 4ft-Miner algorithm. These rules identify compact and interpretable subgroups in the data with elevated prediction confidence and are used to boost the baseline model. The hybrid model is implemented in Python and evaluated on the Telco Customer Churn dataset. The results show that while the overall predictive accuracy of the hybrid and baseline models is comparable, performance improved in uncertain prediction regions, instances where logistic regression predicted class probabilities near 0.5. Notably, some individual rules showed strong generalisation and substantially outperformed the baseline model in their respective subspaces. The design also incorporated techniques to prevent overfitting, including training-only rule mining, support constraints, and rule complexity limits. The findings confirm that boosting predictive models with exception rules is not only feasible but also beneficial in enhancing interpretability and performance. Způsob přístupu: Internet informační management [obor dipl. práce] diplomové práce fd132022 czenas master's theses eczenas boosting logistic regression exception rules predictive model Máša, Petr ISIS:17194 ths Rauch, Jan, 1948- jn20001227021 opn Vysoká škola ekonomická v Praze. Fakulta informatiky a statistiky kn20010709399 dgg https://insis.vse.cz/zp/85786/podrobnosti VŠKP v InSIS https://insis.vse.cz/zp/85786 Hlavní práce https://insis.vse.cz/zp/85786/posudek/vedouci Hodnocení vedoucího https://insis.vse.cz/zp/85786/posudek/oponent/88123 Oponentura https://insis.vse.cz/zp/85786/podrobnosti dc:identifier NEPOSILAT VSKP vse85786 251007 85786