M. Kamber, L. Winstone, W. Gong, S. Cheng, and J. Nan. Generalization and decision tree induction: Efficient classification in data mining. In Proc. of 1997.  Home/Search   Document Details and Download   Summary   Related Articles   Check

.... multi level association rules [15] meta rule guided mining of associations [22] incremental and distributed mining of associations [8, 7] constraint pushing in association mining [10, 27] mining periodicity and similarity in time series data [11, 30] multi level classification and prediction [23, 6], spatial data cube construction [21] spatial association rule mining [24] OLAP mining [12] Weblog mining [31] etc. A data mining system, DBMiner [16, 14] has been constructed with our years of research and development. The system integrates data mining with on line analyt ical processing ....

M. Kamber, L. Winstone, W. Gong, S. Cheng, and J. Nan. Generalization and decision tree induction: Efficient classification in data mining. In Proc. of 1997.

....and data mining systems. One usage is as background knowledge, as in [Han, Cai and Cercone, 1993] Concept hierarchies have also been used in various algorithms for characteristic rule mining [Han, 1995, 1996; Srikant, 1995] multiple level association mining [Han 1995] and classification [Kamber 1997]. What makes our work different is that our concept hierarchy is much larger and more general; it is generated automatically from WordNet over thousands of keywords. Another difference is that it is used directly in guiding the rule induction process. The improvements over the simple ....

M. Kamber, et al, Generalization and Decision Tree Induction: Efficient Classification in Data Mining, Proc. of 1997 Int. Workshop on Research Issues on Data Engineering, Birmingham, England, 1997.

....tree, i.e. building with least branches and attributes, is a NP hard problem [29] Many researchers suggested possible methods to solve this problem. For example, Takeshi Fukuda et al. 17] use a Naive Hand Probing algorithm, with the application of entropy function. Micheline Kamber et al. [32] suggested an optimal pruning algorithm. Sreerama Murthy et al. 39] tested how effective the greedy heuristic decision tree is for the CART and C4.5 decision tree algorithm [42] The C4.5 [42] algorithm is outlined as follows. It starts with a single attribute containing the training examples. ....

Micheline Kamber, Lara Winstone, Wan Gong, Shan Cheng, and Jiawei Han. Generalization and decision tree induction: Efficient classification in data mining. Technical report, Database Systems Research Laboratory, School of Computer Science, Simon Fraser University, B.C., Candada, 1997.

....data axes in view in order to provide an alternative presentation of the data. For example, pivot may be used to transform a 3D cube into a series of 2D planes. The above approaches can be integrated with classification tree induction to provide interactive multi level classification mining [24]. The data cube and knowledge stored in the concept hierarchies are used to induce classification trees at different levels of abstraction. Furthermore, once a classification tree has been derived, the concept hierarchies are used to generalize or specialize individual nodes in the tree, allowing ....

....of data in large datasets, it may not be reasonable to assume that all the tuples within a leaf node always belong to the same class. This problem may be addressed by employing a classification threshold. The classification threshold can be set by a domain expert or statistically determined [24]. Further partitioning of the data subset at a given node is terminated if the percentage of tuples belonging to any given class at that node exceeds this threshold. That is, there exists a major class within the node, which is called majority class (label) at this node. We set the classification ....

[Article contains additional citation context not shown here]

M. Kamber, L. Winstone, W. Gong, S. Cheng, and J. Han. Generalization and decision tree induction: Efficient classification in data mining. In Proc. 1997 Int. Workshop Research Issues on Data Engineering (RIDE'97), pages 111--120, Birmingham, England, April 1997.

....comparator, associator, classifier, and predictor. Several additional data mining modules, including mining from time related data, are at research and development stage. We advise the reader to look at numerous publications that explain DBMiner structure and its data mining techniques [32, 33, 34, 35, 43, 44]. We are aware that data mining is unsupervised learning, and that a user has to CHAPTER 5. IMPLEMENTATION AND EXPERIMENTS 74 direct the discovery process. Therefore, an important issue in designing and developing a data mining system is providing a user with an easy and a straightforward way to ....

....clustered and then, find descriptions for each cluster to determine appropriate marketing strategies. ffl Geo classifier adopts a generalization based decision tree induction method to build a classification tree that classifies a set of relevant data according to one of the nonspatial attributes [44]. The classification tree is displayed and by clicking on any of the nodes of the tree a user highlights corresponding region(s) on the map. For example, one may classify states in the United States according to the median family income in a state. The data mining modules described above use ....

M. Kamber, L. Winstone, W. Gong, S. Cheng, and J. Han. Generalization and decision tree induction: Efficient classification in data mining. In Proc. of 1997 Int. Workshop on Research Issues on Data Engineering (RIDE'97), pages 111--120, Birmingham, England, April 1997.

....non traditional data cubes will enhance the power of data mining. 3. Cube based mining methods. Cube based data mining methods should be the core of the on line analytical mining mechanism. There have been many studies on cube based data mining including concept description [12] classification [16], association [15] prediction [6] clustering [2] etc. Cube based mining may inherit the spirit of relational or transactional data mining methods, such as [3, 8] but may explore many distinct features of cube based multidimensional optimization. More studies are needed on efficient cube based ....

M. Kamber, L. Winstone, W. Gong, S. Cheng, and J. Han. Generalization and decision tree induction: Efficient classification in data mining. In Proc. of 1997 Int. Workshop Research Issues on Data Engineering (RIDE'97), pages 111--120, Birmingham, England, April 1997.

....of concept hierarchy into the attribute oriented induction (AOI) leads AOI to be one of the most successful techniques in data mining. Concept hierarchies have been used in various algorithms such as characteristic rule mining[24] 27] multiple level association mining[26] classification[31] and prediction. Association rule and its initial mining algorithm is proposed by Agrawal, Imielinski and Swami[2] and fast algorithms are reported in Agrawal and Srikant[3] However, they do not consider any concept generalization and only discover patterns using raw data, in other words, the ....

M. Kamber, L. Winstone, W. Gong, S. Cheng and J. Han. Generalization and Decision Tree Induction: Efficient Classification in Data Mining. In Proc. of 1997 Int'l Workshop on Research Issues on Data Engineering (RIDE'97), Birmingham, England, 111-120, 1997.

....the computation. 2. It may use aggregate information. 3. It may use distance based join index [16] to accelerate query processing. Neighborhood index [5] may not be the best for that purpose. 4. It may use concept hierarchies which result in simpler decision trees and faster computations [11]. 5. It uses relevance analysis process to eliminate predicates and attributes that do not contribute to the quality of classification. Complexity analysis. Execution time of the algorithm presented above can be estimated using equations presented below. Time to classify objects without ....

M. Kamber, L. Winstone, W. Gong, S. Cheng, and J. Han. Generalization and Decision Tree Induction: Efficient Classification in Data Mining. Proc. of 1997 Int. Workshop Research Issues on Data Engineering (RIDE'97)(April, Birmingham, England), 1997, pp. 111-120.

....efficiently and effectively, with the following distinct features: 1. It incorporates several interesting data mining techniques, including data cube and OLAP technology [3] attribute oriented induction [6, 9] statistical analysis, progressive deepening for mining multiple level rules [8, 9, 12], and meta rule guided knowledge mining [5, 11] It also implements a wide spectrum of data mining functions including characterization, comparison, association, classification, prediction, and clustering. 2. It performs interactive data mining at multiple levels of abstraction on any ....

M. Kamber, L. Winstone, W. Gong, S. Cheng, and J. Han. Generalization and decision tree induction: Efficient classification in data mining. In Proc. of 1997 Int. Workshop on Research Issues on Data Engineering (RIDE'97), pages 111--120, Birmingham, England, April 1997.

.... multi level association rules [15] meta rule guided mining of associations [22] incremental and distributed mining of associations [8, 7] constraint pushing in association mining [10, 27] mining periodicity and similarity in time series data [11, 30] multi level classification and prediction [23, 6], spatial data cube construction [21] spatial association rule mining [24] OLAP mining [12] Weblog mining [31] etc. A data mining system, DBMiner [16, 14] has been constructed with our years of research and development. The system integrates data mining with on line analytical processing ....

M. Kamber, L. Winstone, W. Gong, S. Cheng, and J. Han. Generalization and decision tree induction: Efficient classification in data mining. In Proc. of 1997 Int. Workshop Research Issues on Data Engineering (RIDE'97), pages 111--120, Birmingham, England, April 1997.

....conditions can be discovered under which the periodic patterns occur. One obvious way to do this is to explicitly express the interested conditions in the selective query to find periodic patterns associated with these conditions. Another way is by using classication rule discovery techniques [26] to describe the discovered patterns by other non time related attributes. Appendix: A AprioriAll Algorithm Agrawal et al. proposed an AprioriAll algorithm for mining sequential patterns in transaction databases[7] The algorithm consists of two phases, a sequence phase for discovery of all ....

M. Kamber, L. Winstone, W. Gong, S. Cheng, and J. Han. Generalization and decision tree induction: Efficient classification in data mining. In Proc. of BIBLIOGRAPHY 95 Int. Workshop on Research Issues on Data Engineering (RIDE'97), pages 111-- 120, Birmingham, England, April 1997.

....a leaf node which holds the class prediction for that sample. ID3 [48, 49] and CART [5] procedures for induction of decision trees have been well established for highly effective classification. Other procedures, such as SLIQ [42] and SPRINT [50] have been developed for very large training sets. [30] proposed an efficient algorithm of decision tress induction. The algorithm has addressed not only the efficiency and scalability issues, but also the innovative multi level classification. Chapter 3 Relevance Analysis 3.1 Motivations Usually, there may be a large number of attributes ....

....Again, we will cross validate the results. Similar results can be observed from Figure 6.5. 6.2.2 Prediction Accuracy Compared with Other Approach ID3 procedure for induction of decision trees has had a significant and continuing influence on machine learning research and applications. In [30], an ID3 based classification algorithm was proposed. This algorithm presented the classification result in a very similar way as that of the proposed prediction approach, i.e. instead of classifying an object into only one class, a class distribution is given. As a result, we shall be able to ....

M. Kamber, L. Winstone, W. Gong, S. Cheng, and J. Han. Generalization and decision tree induction: Efficient classification in data mining. In Proc. of 1997 Int. Workshop on Research Issues on Data Engineering (RIDE'97), pages 111-- 120, Birmingham, England, April 1997.

No context found.

Kamber, M., Winstone, L., Gong, W., Cheng, S., Han, J.: Generalization and Decision Tree Induction: Efficient Classification in Data Mining, Proc. of 1997 Int'l Workshop on Research Issues on Data Engineering (RIDE'97) Birmingham, England, April (1997)

No context found.

M. Kamber, L. Winstone, W. Gong, S. Cheng, and J. Han. Generalization and decision tree induction: Efficient classification in data mining. In Proc. of 1997 Int. Workshop Research Issues on Data Engineering (RIDE'97), pages 111--120, Birmingham, England, April 1997. BIBLIOGRAPHY 95

Online articles have much greater impact   More about CiteSeer.IST   Add search form to your site   Submit documents   Feedback  

CiteSeer.IST - Copyright Penn State and NEC