AUW-CE Mining Algorithms & Dataset Hub

High Utility Co-location Pattern Mining (HUCPM), as an important branch of spatial data mining, aims to extract patterns with utility values that meet or exceed a predefined threshold based on user-defined utility criteria (e.g., cost, profit). However, due to the non-uniformity of spatial distribution, the utility associations between spatial features exhibit significant differences across different regions. As data scale and complexity continue to increase, mining efficiency faces significant challenges. Although various pruning strategies have been proposed to enhance mining efficiency, they cannot adaptively adjust based on the characteristics of the data distribution, making them difficult to apply widely across different datasets. To address these issues, this paper introduces the AUW-CE Miner (Adaptive Utility-Weighted Cross-Entropy Miner), a heuristic algorithm built upon an enhanced cross-entropy framework. By integrating a heuristic search mechanism, the algorithm can quickly converge to potential high utility patterns and effectively reduce redundant computational processes. Moreover, in response to the limitations of conventional cross-entropy methods for HUCPM, four core optimization strategies are designed: optimization of the initial probability distribution to guide the search direction, enhancement of sample diversity to prevent local convergence, dynamic adjustment of sample size to reduce redundant calculations, and incorporation of utility weights to improve the accuracy of probability updates. Experimental results show that the AUW-CE Miner significantly outperforms other algorithms in terms of runtime efficiency, with an average efficiency improvement of up to 56.5\%, demonstrating exceptional mining efficiency and stability.