20.500.12556/RUL-184036 Napovedovanje kakovosti pitne vode z uporabo metod strojnega učenja Predicting the quality of drinking water using machine learning methods Kakovost pitne vode je eden ključnih dejavnikov varovanja javnega zdravja, zato je njeno spremljanje zakonsko predpisano. V Sloveniji se izvaja redni državni monitoring, ki spremlja širok nabor mikrobioloških in kemijskih kazalcev kakovosti. Z naraščajočim obsegom zbranih meritev se odpira vprašanje, ali lahko z metodami strojnega učenja izboljšamo ali dopolnimo tradicionalno presojo skladnosti vzorcev. Namen diplomske naloge je primerjati referenčni pristop z odločitvenimi pravili in metode strojnega učenja pri napovedovanju skladnosti vzorcev pitne vode s predpisanimi kazalci kakovosti. Nalogo smo zasnovali kot binarno klasifikacijo na obsežni množici vzorcev iz slovenskega državnega monitoringa za obdobje 2019–2025. Pristope vrednotimo z metodo časovnega prečnega preverjanja Leave-One-Year-Out, izbor atributov pa opravimo z metodo SHAP. Referenčni pristop z odločitvenimi pravili smo primerjali s štirimi modeli strojnega učenja: Gradient Boosting, Random Forest, LightGBM in večplastni perceptron. Predlagamo tudi dva lastna pristopa: regulativna korekcija metolaklora za pravilno obravnavo razgradnih produktov (ESA, OXA) pri vsoti pesticidov ter dvostopenjski model za boljšo obravnavo redkih kršitev. Rezultati kažejo, da modeli strojnega učenja dosežejo primerljive rezultate z referenčnim pristopom. Predlagana regulativna korekcija metolaklora bistveno zmanjša število napak referenčnega pristopa. Drinking water quality is a key public health factor and its monitoring is mandated by law. In Slovenia, a regular national monitoring programme tracks a wide range of microbiological and chemical quality indicators. With the growing volume of measurements collected, the question arises whether machine learning methods can improve or complement the traditional compliance assessment. The goal of this thesis is to compare a rule-based reference approach and machine learning methods for predicting the compliance of drinking water samples with prescribed quality indicators. The task is formulated as a binary classification problem on a large dataset of samples from the Slovenian national monitoring programme for the period 2019-2025. Approaches are evaluated using Leave-One-Year-Out time-series cross-validation, with feature selection performed using the SHAP method. We compare the rule-based reference approach against four machine learning models: Gradient Boosting, Random Forest, LightGBM, and a Multi-Layer Perceptron. Two novel approaches are also proposed: a regulatory correction for metolachlor aimed at the correct treatment of degradation products (ESA, OXA) in the pesticide sum assessment, and a two-step model for improved handling of rare violations. Results show that machine learning models achieve comparable performance to the reference approach. The proposed regulatory correction for metolachlor substantially reduces the number of errors in the reference approach. strojno učenje napovedovanje kakovost pitne vode machine learning prediction drinking water quality true false false Slovenski jezik Angleški jezik Diplomsko delo/naloga 2026-06-24 11:05:01 2026-06-24 11:05:11 2026-07-23 08:59:50 0000-00-00 00:00:00 2026 0 0 0000-00-00 NiDoloceno NiDoloceno NiDoloceno 0000-00-00 0000-00-00 0000-00-00 38950 285779203 Sovic_Aljaz_-_Napovedovanje_kakovosti_pitne_vode_z_uporabo_metod_strojnega_ucenja.pdf Sovic_Aljaz_-_Napovedovanje_kakovosti_pitne_vode_z_uporabo_metod_strojnega_ucenja.pdf 1 EA7D21494065408DCC3B540FBA660943 3e1f83731f826570001ba6b115e426d0daab75d8f117cddf5857500d97b9b193 b9b0a129-6fab-11f1-9b0d-0050569b8976 https://repozitorij.uni-lj.si/Dokument.php?lang=slv&id=236483 Fakulteta za računalništvo in informatiko 0 0 0