• Share
  • Email
  • Embed
  • Like
  • Save
  • Private Content
Av24317320
 

Av24317320

on

  • 167 views

 

Statistics

Views

Total Views
167
Views on SlideShare
167
Embed Views
0

Actions

Likes
0
Downloads
0
Comments
0

0 Embeds 0

No embeds

Accessibility

Categories

Upload Details

Uploaded via as Adobe PDF

Usage Rights

© All Rights Reserved

Report content

Flagged as inappropriate Flag as inappropriate
Flag as inappropriate

Select your reason for flagging this presentation as inappropriate.

Cancel
  • Full Name Full Name Comment goes here.
    Are you sure you want to
    Your message goes here
    Processing…
Post Comment
Edit your comment

    Av24317320 Av24317320 Document Transcript

    • S. Thiripura Sundari, A. Padmapriya / International Journal of Engineering Research and Applications (IJERA) ISSN: 2248-9622 www.ijera.com Vol. 2, Issue 4, July-August 2012, pp.317-320 Data Mining Application-Usage of Visualizing Association Rules in CRM System S. Thiripura Sundari*, Dr.A. Padmapriya** *(Department of Computer Science and Engg, AlagappaUniversity, Karaikudi-630 003) ** (Department of Computer Science and Engg, AlagappaUniversity, Karaikudi-630 003)ABSTRACT Data Mining, simply called DM, is a kind mining is the key to the successful implementation ofof method for data analyses at deep level. Data CRM. And to use data mining techniques to customerMining can be described as this: It is, in light of the analysis to help enterprise to better implement CRMset target of an enterprise, an effective and and develop customized marketing strategies. Thisadvanced method in probing and analyzing huge kind of customer relationship management, customer-numbers of enterprise data and revealing the centric business strategy, is based on informationunknown regularities or verifying the known technology. It restructures the work processes in orderdisciplines and model them better. Association rule to give businesses better customer communicationmining is one of the most popular data mining skills, to maximize customer profitability.methods. However, mining association rules oftenresults in a very large number of found rules, The main objective of its management is toleaving the analyst with the task to go through all increase efficiency, expand markets and retainthe rules and discover interesting ones. Sifting customers. Data mining is data processing in a high-manually through large sets of rules is time level process, which from the data sets to identify inconsuming and strenuous. Visualization has a long order to model to represent knowledge. The so-calledhistory of making large amounts of data better high-level process is a multi-step processing, multi-accessible using techniques like selecting and step interaction between the repeated adjustments, tozooming. In this paper we introduces the form a spiral process.Visualizing Association Rules using CRM baseddata mining and the concept of Clustering The knowledge of mining expresses asAnalysis, and then started from Customer Concepts, Rules, Regularities and other forms.relationship management’s core values-Customer Clustering Analysis is a widely used data miningValue, deeply expound the meaning of customer analytical tools. After clustering, the same types ofvalue and it is in a key position in the customer data together as much as possible, while the separationrelationship management. of different types of data as possible. The enterprise customer classification, through the customer’s buyingKeywords - Customer Relationship Management, behavior, consumption habits, and the background toData mining identify the major clients, customers, and tap potential customers. There are many clustering algorithms, thisI. INTRODUCTION means using K-means algorithm. K-means algorithm Many organizations are generating a large is a specific means: for a given number of classes K,amount of transaction data on a daily basis. the n objects assigned to a class K to go, makingEnterprises must rely on customers to achieve within-class similarity between objects the most, whileprofitable. In the fierce market competition, in order to the similarity between the smallest class.maintain superiority and long-term and stabledevelopment, we must attach importance to customer II. BACKGROUND OF THE STUDYrelationship management. And only continued to gain Association rule mining is a well-known dataand maintain valuable customers and achieve customer mining task for discovering association rules betweendemand for personalized, in-depth excavation in order items in a dataset. It has been successfully applied toto achieve the corporation-customer win-win. This different domains especially for business applications.paper describes in depth the importance of customer However, the mined rules rely heavily on humanvalue for the company, noting that customer value interpretation in order to infer their semantic meanings. In this paper, we mine a new kind of 317 | P a g e
    • S. Thiripura Sundari, A. Padmapriya / International Journal of Engineering Research and Applications (IJERA) ISSN: 2248-9622 www.ijera.com Vol. 2, Issue 4, July-August 2012, pp.317-320association rules, called conceptual association rules, antecedents and consequents in rules. Graph-basedwhich imply the relationships between concepts. visualization offers a very clear representation of rulesConceptual association rules can convey more but they tend to easily become cluttered and thus aresemantic meanings than those classical association only viable for very small sets of rules.rules. Parallel coordinates plots are designed to In Data Mining, the usefulness of association visualize multidimensional data where each dimensionrules is strongly limited by the huge amount of is displayed separately on the x-axis and the y-axis isdelivered rules. In this paper we propose a new shared. Each data point is represented by a lineapproach to prune and filter discovered rules. Using connecting the values for each dimension.Domain Ontologies, we strengthen the integration ofuser knowledge in the post-processing task. Double decker plots use only a singleFurthermore, an interactive and iterative framework is horizontal split. Introduced double decker plots todesigned to assist the user along the analyzing task. visualize a single association rule.On the one hand, we represent user domain knowledgeusing a Domain Ontology over database. 1.2. Drawbacks in Existing Method Two key plot - There is not enough space in K-means clustering is a popular clustering the plot the display the labels of the items in the rules.algorithm based on the partition of data. However, this still makes exploring a large set of rules time-there are some shortcomings of it, such as its requiring consuming.a user to give out the number of clusters at first, and itssensitiveness to initial conditions, and its easily getting This visualization is powerful for analyzing ato the trap of a local solution et cetera. The global K- single rule; it cannot be used for large sets of rules.means algorithm proposed by Likas et al is anincremental approach to clustering that dynamically III. PROPOSED CRM SYSTEM USINGadds one cluster center at a time through a VISUALIZING ASSOCIATION RULESdeterministic global search procedure consisting of N Matrix-based visualization techniques(with N being the size of the data set) runs of the K- organize the antecedent and consequent item sets onmeans algorithm from suitable initial positions. the x and y-axes, respectively. Matrix-based visualization is limited in the number of rules it can Data mining technology has emerged as a visualize effectively since large sets of rules typicallymeans for identifying patterns and trends from large also have large sets of unique antecedents/consequentsquantities of data. Mining encompasses various Grouped Matrix-based Visualization- Groupsalgorithms such as clustering, classification, and of rules are presented by aggregating rows/columns ofassociation rule mining. In this paper we take the matrix. The groups are nested and organizedadvantage of the genetic algorithm (GA) designed hierarchically allowing the user to explore themspecifically for discovering association rules. We interactively by zooming into groups. This paperpropose a novel spatial mining algorithm, called attempts to use data mining - clustering method,ARMNGA(Association Rules Mining in Novel combined with customer life cycle assessment systemGenetic Algorithm), the ARMNGA algorithm avoids for the value of customer segmentation. Through datagenerating impossible candidates, and therefore is mining technology can effectively support the use ofmore efficient in terms of the execution time. appropriate evaluation methods to make up a simple application of conventional evaluation methods lack of1.1. Existing Methodology customer value. A straight-forward visualization ofassociation rules is to use a scatter plot with two 1.1. The Advantages are listed belowinterest measures (typically support and confidence) Matrix-based visualization is very limitedon the axes. This is introduced a special version of a when facing large rule sets.scatter plot called Two-key plot. Here support andconfidence are used for the x and y-axes and the color Since minimum support used in associationof the points is used to indicate “order,” i.e., the rule mining leads in addition to relatively short rulesnumber of items contained in the rule. resulting in extremely sparse data. Graph-based techniques visualize association IV. PERFORMANCE ANALYSISrules using vertices and directed edges where vertices In Data Mining, the subject to be studied istypically represent items or itemsets and edges connect the core of the whole process. It drives the whole 318 | P a g e
    • S. Thiripura Sundari, A. Padmapriya / International Journal of Engineering Research and Applications (IJERA) ISSN: 2248-9622 www.ijera.com Vol. 2, Issue 4, July-August 2012, pp.317-320process and is the basis of examining the final results policy. The pie chart analysis all the input values andand directing the analyst to complete mining. calculates the position of the customer using Naive Bayesian Classification after shows the result. So we easily get which one will get insurance pother than deep analysis. V. CONCLUSION The method addresses the problem that sets of mined association rules are typically very large by grouping antecedents and allowing the user to interactively explore a hierarchy of nested groups. Coloring and the position of elements in the plot guide the user automatically to the most interesting groups/rules. Finally, the ability to interpret the grouped matrix-based visualization can be easily acquired since it is based on the easy to understand concepts of matrix-based visualization of association rules and grouping. Customer value analysis is the enterprise implementation of customer relationship management core. For companies to develop customized marketing strategies, it is necessary to segment Fig. 1 Graph bar chart customer groups. This paper analyzes the shortcoming of traditional method of evaluation of customer value.1.1. Clustering or Grouping of Data Through the introduction of data mining-clustering The data collected using any one of the algorithm to improve the traditional methods ofprediction or description method and that data are evaluation, to enable enterprises to more accuratestored as a record. The collected records called as a positioning for target groups with limited enterprisetraining set that data are called as training set data. resources to the implementation of targetedEach record contains a set of attributes; one of the personalized marketing policy for the enterprise-toattributes is class. Find a model for class attribute as a help customers achieve a win-win.function of the values of other attributes. Unlike methods based on a single clustering Attribute values are numbers or symbols of the data, the approach computes the probability thatassigned to an attribute. Same attribute can be mapped two genes belong to the same cluster while taking intoto different attribute values. Different attributes can be account the main sources of model uncertainty,mapped to the same set of values. including the number of clusters and the location of clusters.1.2. Data classification The method used here to obtain a predictive .Referencesmodel, and to answer the questions above, is naive [1] Abd. Manan Ahmad, AG. Noorajis Ag.Bayesian learning. Boosting was tried also, and some Nordin, Emrul Hamide Md. Saaim, Fairolderived attributes were added to the training set. After Samaon and Mohd Danial Ibrahim, “Antwo important derived attributes were added, boosting architecture design of the intelligent agentdid not give any significant increase in accuracy on a for speech recognition and translation,” invalidation set taken from the training set, and neither IEEE, 2004.did adding more derived attributes. Therefore, the [2] Bala, J., Weng, Y., Wiliams, A.I., Gogia,results here do not use boosting, and only use two B. K. and H. K. Lesser, “Applications ofderived attributes. distributed mining techniques for1.3. Analysis with graph bar chart knowledge discovery in dispersed sensory The results are shown using Graph chart. data”, in 7th Joint Conference onHere we have used 2 numerical parameters that Information Sciences- JCIS,2003.parameters are 0 and 1.The result shows 1 means the [3] Bose, R. and V. Sugumaran, “Applicationcustomer ready to get insurance policy. The result of intelligent agent technology forshows 0 means the customer not ready to get insurance managerial data analysis and mining,” in The DATABASE for Advance in 319 | P a g e
    • S. Thiripura Sundari, A. Padmapriya / International Journal of Engineering Research and Applications (IJERA) ISSN: 2248-9622 www.ijera.com Vol. 2, Issue 4, July-August 2012, pp.317-320 Information Systems, 3 (1): pp. 77-94, [8] Ping Luo, Qing He, Rui Huang, Fen Lin 1999. and Zhongzhi Shi, “Execution engine of [4] Freitas, A. Data mining and knowledge meta-learning system for kdd in multi- discovery with evolutionary algorithms, agent environment,” in AIS-ADM 2005: Natural Computing Series, Springer: pp. 149-160, 2005M. Berlin, 2002. [9] Witten, I. H. and E. Frank. Data mining: [5] Han, J. and Kamber, M. Data Mining : practical machine learning tools and Concept and Techniques, Morgan techniques with java implementation. Kaufmann Publisher: San Francisco, USA, Morgan Kaufmann Publishers, San 2001. Francisco,USA,2000.http:/www.cs.waikato [6] Kehagias, D., Chatzidimitriou, K.C., .ac.nz/ml/weka/. Symeonidis, A.L. and P.A. Mitkas, [10] Fayyad, U., Piatetsky-Shapiro, G. and P. “Information agents cooperating with Smyth. The KDD process for extracting heterogeneous data sources for customer- useful knowledge from volumes of data. order management,” in ACM Symposium Communications of the ACM, 39(11): 27- on Applied Computing: pp. 52-57, 2004. 34, 1996. [7] Lourenço, A., Gonçalves, J. and O. Belo, “Agent-based knowledge extraction services inside enterprise data warehousing systems environments,” IEEE: pp. 887- 891, 2001.S.Thiripura Sundari, a very enthusiastic and energetic Dr.A.Padmapriya is the Assistant Professor in thescholar in M.Phil degree in Computer Science. Her Department of Computer Science and Engineering,research interest is in the field of data mining. It will Alagappa University, Karaikudi. She has been a memberfurther be enhanced in future. She is equipping herself to of IACSIT. She has presented a number of papers inthat by attending National Conferences and by both National and International Conferences.publishing papers in International Journals. 320 | P a g e