Water pipeline failures in urban networks are a significant source of non-revenue water, service disruptions, and high maintenance costs. This study develops a machine learning model to predict pipeline failure probabilities and inform risk-based maintenance strategies. Trained on real-world assets and geospatial data from 2010 to 2025, the model incorporates standard pipe attributes – such as material, age, diameter, network type, and maintenance history – alongside spatially derived indicators of the surrounding infrastructure. Notably, it quantifies the predictive impact of adjacent infrastructure systems, including electricity grids, gas pipelines, district heating, sewage systems, and roads, utilizing spatial buffering and overlay techniques. Several of these cross-utility features, particularly road category, electricity voltage, and sewer type, showed meaningful predictive importance, reflecting their indirect but consistent influence on the risk of pipe failure. The ML model, built with the XGBoost algorithm and validated through stratified K-fold cross-validation, achieved high performance (ROC AUC: 0.9102, recall: 0.7750, accuracy: 0.8750). Despite lower precision due to class imbalance, the F1 score (0.2261) and LogLoss (0.2500) confirm its reliability. This study introduces a novel, spatially enriched approach to failure prediction, advancing urban infrastructure management through context-aware, data-driven insights.

This study examines the determinants of apartment prices in 17 post-WWII multi-family housing estates in Ljubljana, Slovenia, constructed between 1947 and 1986. Using 1973 verified transactions from 2020 to 2025, the analysis evaluates spatial, structural, environmental, and accessibility-related variables through a combination of statistical and machine learning techniques. A hedonic price model based on ordinary least squares (OLS) demonstrates modest explanatory power (R² = 0.171), identifying local market reference prices, floor level, noise exposure, and window renovation as significant predictors. In contrast, seven machine learning models—Random Forest, XGBoost, and Gradient Boosting Machines (GBMs), including optimized versions—achieve notably higher predictive accuracy. The best-performing model, GBM with Randomized Search CV, explains 59.6% of price variability (R² = 0.5957), with minimal prediction error (MAE = 0.03). Feature importance analysis confirms the dominant role of localized price references and structural indicators, while environmental and accessibility variables contribute variably. In addition, three clustering methods (Ward, k-means, and HDBSCAN) are employed to identify typological groups of neighborhoods. While Ward’s and k-means methods consistently identify four robust clusters, HDBSCAN captures greater internal heterogeneity, suggesting five distinct groups and detecting outlier neighborhoods. The integrated approach enhances understanding of spatial housing price dynamics and supports data-driven valuation, urban policy, and regeneration strategies for post-WWII housing estates in Central and Eastern European contexts.