Off-campus UNL users: To download campus access dissertations, please use the following link to log into our proxy server with your NU ID and password. When you are done browsing please remember to return to this page and log out.
Non-UNL users: Please talk to your librarian about requesting this dissertation through interlibrary loan.
Temporal data mining methodologies in a geo-spatial decision support system
In this dissertation, temporal data mining methodologies are developed to facilitate knowledge discovery in the framework of a distributed Geo-spatial Decision Support System (GDSS), with a focus on drought risk management. In this process, climatic data are collected from a variety of sources at weather stations. However, there are two kinds of missing (or incomplete) data. First, data are partially missing because of temporary malfunction or unavailability of equipment. Imputation methods based on clustering and soft computing techniques are developed to solve this missing data problem. Second, some locations do not have local observed data due to cost, physical, or technical considerations. To generate association rules for these un-sampled locations, three spatial interpolation models are developed and integrated into the temporal data mining process. ^ After data preparation and preprocessing, we look more closely at the temporal property of time series data. Because a periodic pattern indicates something persistent and predictable, it is important to identify and characterize the periodicity. In this dissertation, an approach for mining partial periodic association rules in temporal databases is discussed. This approach allows the discovery of periodic episodes such that the events in an episode are not constrained by a fixed order. Moreover, this approach treats the antecedent and consequent of a rule separately and allows time lag between them. Thus, rules discovered are useful for prediction. ^ Additionally, droughts occur infrequently by nature. To facilitate drought risk management, it is important to discover infrequent episodes from multiple data sequences. In this dissertation, an algorithm is developed for the discovery of infrequent episodes with a combination of bottom-up and top-down scanning schema. The information sharing between bottom-up and top-down scanning helps prune candidate episodes, and thus, efficiently find infrequent episodes that are interesting to users. ^ Overall, the objective of this research is to enhance the body of work in the area of temporal data mining to enable knowledge discovery in the context of a GDSS. ^
Li, Dan, "Temporal data mining methodologies in a geo-spatial decision support system" (2005). ETD collection for University of Nebraska - Lincoln. AAI3180804.