R. Rastogi and K. Shim. Mining optimized association rules with categorical and numeric attributes. IEEE TKDE, 14(1), January 2002.  Home/Search   Document Details and Download   Summary   Related Articles   Check

....subsequent discussion we shall focus on the first step. There has been a lot of research work in association rule mining. Although the first applications were found in supermarket data, the technique has been extended to work on numerical data and categorical data in more conventional databases [SA96, RS98], some researchers have noted the importance of association rule mining in relation to relational databases [STA00] Tools for association rule mining are now found in major products such as IBM s Intelligent Miner, and SPSS s Clementine. However, research and development so far have mostly ....

....the same attribute will not appear in two different tables. If initially two tables have some common attributes, renaming can make them different. We assume that attributes take categorical values. Numerical values can be partitioned into ranges, and hence be transformed to categorical values [RS98]. The set of values for an attribute is called the domain of the attribute. Conceptually, we can view the dimension table in terms of a binary representation, where we have one binary attribute (or we call an item ) corresponding to one attribute value pair in the original dimension table. We ....

R. Rastogi and K. Shim, Mining Optimized Association Rules with Categorical and Numeric Attributes, Proc. of International Conference on Data Engineering (ICDE), pp 503-512, 1998.

....we generate the association rules. The first step is more difficult and is shown to be NP hard. In our subsequent discussion we shall focus on the first step. The techniques in association rule mining has been extended to work on numerical data and categorical data in more conventional databases [SA96, RS98], some researchers have noted the importance of association rule mining in relation to relational databases [STA00] Tools for association rule mining are now found in major products such as IBM s Intelligent Miner, and SPSS s Clementine. In real databases, typically a number of tables will be ....

....that the attributes in the dimension tables are unique. If initially two tables have some common attributes, renaming can make them different. We assume that attributes take categorical values. Numerical values can be partitioned into ranges, and hence be transformed to categorical values [RS98]. The set of values for an attribute is called the domain of the attribute. Conceptually, we can view the dimension table in terms of a binary representation, where we have one binary attribute (or we call an item ) corresponding to one attribute value pair in the original dimension table. We ....

R. Rastogi and K. Shim, Mining Optimized Association Rules with Categorical and Numeric Attributes, Proc. of International Conference on Data Engineering (ICDE), pp 503-512, 1998.

....concept of az interesting data set. This has led to the notion of associaiion rules [1, 2] and their variants, e.g. generalized association rules [24, 12] correlation rules [4] and causal structures [23] ratio rules [18] quantitative association rules [25, 31, and optimized association rules [6, 7, 21]. Association rules define a specific type of hypothesis and the goal of the proposed algorithms is to find in stantiations for attributes to derive rules that make the hypothesis more specific mid allow the user to confirm or reject them. In its most basic definition a rule is m expression of ....

....to large data sets. In addition, it poses the need for additional tools for extracting the most interesting patterns. To alleviate some of the problems that arise when generating all possible rules that satisfy a certain threshold, the class of optimization based rule miner [7] was proposed. In [21] the framework of optimized association rules, described in [6, 7] was extended in several ways. How ever, the rules obtained suffer from two problems. They pre specify the vm iable in the hypothesis, and their consequence is a categorical variable, rather thal numeric. 1.2 Data Visualization ....

R. Rastogi and K. Shim. Mining optimized association rules with categorical and numeric attributes. In Proc. of the ldth Int'l Conf. on Data Engineering, pages 503-512, 1998.

....concept of an interesting data set. This has led to the notion of association rules [1, 2] and their variants, e.g. generalized association rules [29, 15] correlation rules [4] and causal structures [28] ratio rules [21] quantitative association rules [30, 3] and optimized association rules [9, 10, 26]. Association rules define a specific type of an hypothesis and the goal of the proposed algorithms is to find instantiations for attributes to derive rules that make the hypothesis more specific and allow the user to confirm or reject them. In its most basic definition a rule is an expression of ....

....of optimization based rule miner [10] was proposed. The purpose here is to identify only rules that are optimized according to some measure of interest. This approach is usually preferred when the constraint based approach generates an excessive amount of rules and results in poor performance. In [26] the framework of optimized association rules, described in [9, 10] was extended in several ways. However, the rules obtained suffer from two problems. They pre specify the variable in the hypothesis, and their consequence is a categorical variable, rather than numeric. 1.2 Data Visualization ....

R. Rastogi and K. Shim. Mining optimized association rules with categorical and numeric attributes. In Proc. of the 14th Int'l Conf. on Data Engineering, pages 503--512, 1998.

....which exist and are being built by many organizations. Association rule mining [2] is perhaps the most researched problem of the three. Extensions to the problem include the inclusion of the effect of time on association rules [6] 11] and the use of continuous numeric and categorical attributes [12]. Mining sequential patterns is explored in [4] Therein, a pattern is a sequence of events attributed to an entity, such as items purchased by a customer. Like association rule mining, 4] reduces the search space by using knowledge from size k patterns when looking for size k 1 patterns. ....

....range of observations, thereby coalescing similar values. This, of course, assumes that such groups exist in the data. The output of the regression tree methods in [8] can be used to segment continuous values into meaningful subgroups. The numeric ranges chosen for attributes in output from using [12] can also be utilized for segmentation. In the absence of such patterns, another method is to statistically separate the continuous data by using standard deviation and average metrics. This is the approach used in this paper for transforming the Oracle performance data. Another method is to ....

Rastogi, R. and Shim, K. Mining Optimized Association Rules with Categorical and Numeric Attributes. IEEE Data Engineering, 1998.

....mining (Agrawal, Imielinski et al. 1993) is perhaps the most researched problem of the three. Extensions to the problem include the inclusion of the e ect of time on association rules (Chakrabarti et al. 1998) Ramaswamy et al. 1998) and the use of continuous numeric and categorical attributes (Rastogi, Shim, 1998). Mining sequential patterns is explored in (Agrawal, Srikant et al. 1995) Therein, a pattern is a sequence of events attributed to an entity, such as items purchased by a customer. Like association rule mining, Agrawal, Srikant et al. 1995) reduces the search space by using knowledge from ....

....thereby coalescing similar values. The output of the regression tree methods in (Morimoto et al. 1997) can be used to partition continuous values into meaningful subgroups. This, of course, assumes that such groups exist in the data. The numeric ranges chosen for attributes in output from using (Rastogi, Shim, 1998) can also be utilized for segmentation. In the absence of such patterns, another method is to statistically separate the continuous data by using standard deviation and average metrics. This is the approach used in this paper for transforming the computer system performance data used in our ....

Rastogi, R. and Shim, K. (1998). Mining Optimized Association Rules with Categorical and Numeric Attributes. IEEE Data Engineering.

....or support values of the rules are maximized. However, their schemes were only suitable for a single optimal region. Rastogi and Shim thus extended the problem for more than one optimal regions, and showed that the problem was NP hard even for the case of one uninstantiated numeric attribute [28, 29]. Some works also used fuzzy set theory and data mining technology to solve classification problems [7, 23] 7 In this paper, we use fuzzy set concepts to mine association rules from transactions with quantitative attributes. The mined rules are expressed in linguistic terms, which are more ....

R. Rastogi and K. Shim, "Mining optimized association rules with categorical and numeric attributes," The 14th IEEE International Conference on Data Engineering, Orlando, 1998, pp. 503-512.

....the values of a quantitative attribute into non overlapping partitions optimally. Fukuda et al. FMMT96] introduced optimized association rules, which tries to find the partitioning of values of numerical attributes to maximize the support or confidence. The same concept was also investigated in [RS98] for numerical and categorical attributes. Wang et al. WTL98] proposed an interestingness based interval merger for combining different intervals CHAPTER 2. A SURVEY IN ASSOCIATION RULES 26 to one in order to maximize the interestingness of a rule. In another study related to numerical ....

....algorithms for discovering frequent episodes. 2.6.5 Periodical Rules In a sequence of data, association rules may reveal periodical properties. Moreover, some of the rules may have enough support in a smaller time period even it does not have enough support in the global database. Ozden et al. ORS98] introduce cyclic association rules, which are the rules that have the specified confidence and support in regular time intervals. One such rule states that People buy newspapers along with milk every Sunday . Instead of finding the rules at each time point, and then attempting to generate ....

[Article contains additional citation context not shown here]

Rajeev Rastogi and Kyuseok Shim. Mining optimized association rules with categorical and numerical attributes. In Proceedings of 14 th Intl. Conf. on Data Engineering (ICDE'98), pages 503--512, Orlando, Florida, USA, February 1998.

....then mapping this problem to the traditional association rule mining problem. It only utilizes the support threshold to prune the search space. In contrast, our proposed algorithm utilizes all thresholds for pruning the search space, and the effects are shown in Section 5.2. Fuk96a] Fuk96b] [Ras98] [Ras99] focused on mining so called optimized association rules. One or more numerical attributes are allowed on the left hand side of the rule while the right hand side is restricted to categorical attributes. We allow categorical or numerical variables to be freely mixed 2 . Most research in ....

....schemes to determine values for variable l and u such that one of confidence or support of the rule is maximized while the other satisfies the requirement. Fuk96b] extended the results in [Fuk96a] to the case in which rules contain two uninstantiated numerical attributes on the left hand side. [Ras98] again generalized the problem and proposed algorithms to handle more than two uninstantiated attributes on the left hand side which can be either categorical or numerical. Ras99] further generalized optimized support association rule problem by permitting rules to contain disjunctions over ....

R. Rastogi and K. Shim. Mining optimized association rules with categorical and numerical attributes. Proc. 14th Intl. Conf. on Data Engineering (ICDE), 1998.

No context found.

R. Rastogi and K. Shim. Mining optimized association rules with categorical and numeric attributes. IEEE TKDE, 14(1), January 2002.

No context found.

R. Rastogi and K. Shim. Mining optimized association rules with categorical and numeric attributes. IEEE Transactions on Knowledge and Data Engineering, 14(1), January 2002.

No context found.

Rajeev Rastogi and Kyuseok Shim. Mining optimized association rules with categorical and numeric attributes. In Proceedings of the 14th International Conference on Data Engineering, pages 503--512. IEEE Computer Society, 1998.

No context found.

R. Rastogi and K. Shim. Mining optimized association rules with categorical and numeric attributes. IEEE TKDE, 14(1), January 2002.

No context found.

R. Rastogi and K. Shim, "Mining Optimized Association Rules with Categorical and Numeric Attributes," Proceedings of the IEEE International Conference on Data Engineering, Orlando, FL, 1998, pp. 503-512.

No context found.

Rastogi, R. and Shim, K. 1998. "Mining optimized association rules with categorical and numeric attributes." Proceedings of lEEE International Conference on Data Engineering (ICDE-98), pp. 503-512, 1998.

No context found.

Rastogi, R. and Shim, K. "Mining optimized association rules with categorical and numeric attributes." ICDE --98.

No context found.

Rastogi R., and Shim K. "Mining Optimized Association Rules with Categorical and Numeric Attributes." Proceedings of the International Conference on Data Engineering, Orlando, Florida, February 1998.

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