For The Large Amount Of Data Mining Algorithms Differential Grid Space Co-location Model

Recently, with the rapid growth of the amount of data in the information age, it becomes increasingly important to mine useful information from a massive data. Spatial co-location pattern represent a subset of spatial features and whose instances are frequently associated in space. It is an important task of spatial data mining for spatial co-location pattern mining, and we have done lots of research on it. However, the efficiencies of time and space are low for traditional mining algorithms of determination data and uncertain data, so how to work out some mining algorithms with high efficiencies of time and space, which has been our research goal and direction.In fact, researchers have proposed many improved algorithms, but the existing algorithms are still very sensitive to the distribution intensity of instances and the amount of data, the consumption of algorithms’time and space resources will suddenly increase whether the intensity of instances or the amount of data increase slightly. What’s more, the distribution of instances tends to be more uneven and localized dense, and the amount of data is very large in real life. In this case, the existing algorithms will consume lots of time and space resources to compute in the local intensive area, while which is often unnecessary. Therefore, if we can decrease the complexity of algorithms’ time and space, we will solve the problem of algorithms’efficiency fundamentally, and greatly improve the overall efficiency of the mining algorithms, which will have major significance and application prospects.In order to effectively improve the efficiency of existing algorithms’time and space, especially for the key issues whose instances have a high-density distribution in the local or global area, we propose a grid differential algorithm. Experimental results show that the algorithm can effectively solve related problems in real life with an excellent efficiency of time and space. Details are as follows:Firstly, I analyze the research status quo of co-location pattern mining, summarize its research contents and research results, and introduce the related definitions of spatial co-location pattern mining in this paper.Secondly, I introduce the existing problems of traditional algorithms and realistic background, basic ideas and theory, and implementation of grid differential algorithm we proposed. Thirdly, two grid differential algorithms are designed based on4differential gird and9differential gird, and its basic idea and implementation are introduced in detail. I also make in-depth theoretical analysis, comparison, and demonstration for the traditional algorithm (Join-Base algorithm), the multi-resolution pruning (grid) algorithm, and the grid differential algorithm we proposed in terms of the complexity of algorithms’ time and space.Fourthly, I respectively verify the high efficiency of the grid differential algorithm with experiments on synthetic and real data, analyze the influence on algorithm performance of each parameter and the distribution of instances, and verify the advantages and disadvantages of grid differential algorithm.Fifthly, the co-location pattern mining based on grid differential algorithm is applied to Three Parallel Rivers Project, digging out the interdependence relationship of plants in three parallel rivers region, which provide decision support for users.Sixthly, the intelligent theories and methods of algorithm selection are proposed, which can help users to make scientific choices through the trade-offs between the traditional algorithms and grid differential algorithm we proposed according to this kind of theories and methods.Finally, this paper briefly summarizes the whole work, and points out the disadvantages and future research directions.