Colocation pattern mining identifies spatial features that frequently occur together in a geographical area. Colocation mining is key for identifying spatial patterns and relationships between objects or events, aiding in fields like urban planning, environmental monitoring, and marketing. Existing approaches focus on computational challenges, but few address how pattern interestingness can vary across regions due to local contexts. We propose a novel, memory-efficient map-based approach that reduces memory usage by 70% compared to array-based approaches while identifying sub-regional and regional patterns. Our preliminary evaluation of two real-world case study regions using the Global Terrorism Database demonstrates promising results, highlighting how pattern interestingness varies across regions and showcasing the memory efficiency of our approach. The source code for our work can be accessed at https://bit.ly/3Ox6DZh .