International Journal of Computer Applications (0975 – 8887) Volume 4 – No.5, July 2010 33 Using Associative Classifiers for Predictive Analysis in Health Care Data Mining Sunita Soni O.P.Vyas Associate Professor Professor Bhilai Institute of Technology, Indian Institute of Information Durg-491 001, Chhattisgarh, India Technology, Allahabad- 211012 (U.P.), India ABSTRACT Association rule mining is one of the most important and well researched techniques of data mining for descriptive task, initially used for market basket analysis. It finds all the rules existing in the transactional database that satisfy some minimum support and minimum confidence constraints. Classification using Association rule mining is another major Predictive analysis technique that aims to discover a small set of rule in the database that forms an accurate classifier. In this paper, we introduce the combined approach that integrates association rule mining and classification rule mining called Associative Classification (AC). This is new classification approach. The integration is done by focusing on mining a special subset of association rules called classification association rule (CAR). And then classification is being performed using these CAR. Using association rule mining for constructing classification systems is a promising approach. Given the readability of the associative classifiers, they are especially fit to applications were the model may assist domain experts in their decisions. Medical field is a good example was such applications may appear. Consider an example were a physician has to examine a patient. There is a considerable amount of information associated with the patient (e.g. personal data, medical tests, etc.). A classification system can assist the physician in this process. The system can predict if the patient is likely to have a certain disease or present incompatibility with some treatments. Considering the output of the classification model, the physician can make a better decision on the treatment to be applied to this patient. There are many associative classification approaches that have been proposed recently such as CBA, CMAR, CPAR and MCAR and MMAC. Also Combining the Advanced association rule mining with classifiers gives a new type of Associative classifiers with small refinement in the definition of support and confidence that satisfies the validation of downward closure property. We will discuss advanced associative classifiers being proposed in recent years to provide better accuracy as compare to traditional Classifiers. Keywords Associative Classifiers, CBA, CMAR, CPAR, MCAR 1. INTRODUCTION Data mining is a process, which involves the application of specific algorithms for extracting patterns (models) from data. New knowledge may be obtained in the process while eliminating one of the largest costs, viz., data collection. Medical data, for example, often exists in vast quantities in an unstructured format. A new predictive modeling approach known as associative classification, integrating association Mining and classification inside into single system is being discussed as better alternative for predictive analytics [3]. Some of the classification techniques presented are CBA[10], CMAR[9], CPAR[8]. As discussed in [10] it achieves higher classification accuracy than do traditional classification approaches such as C4.5, FOIL, RIPPER. According to [10] these traditional classifiers are faster but in many cases accuracy is not so high. Moreover many of the rules found by associative classification method cannot be discovered by traditional classification algorithm. Given the readability of the associative classifiers, they are especially fit to applications were the model may assist domain experts in their decisions. Medical field is a good example were such applications may appear. Let us consider an example were a physician has to examine a patient. There is a considerable amount of information associated with the patient (e.g. personal data, medical tests, etc.). A classification system can assist the physician in this process. The system can predict if the patient is likely to have a certain disease or present incompatibility with some treatments. Considering the output of the classification model, the physician can make a better decision on the treatment to be applied to this patient [6]. The rest of the paper is organized as follows. The concept of Associative Classifiers (AC) is being discussed in section 2. In section 3 the advanced AC that are recently proposed have been introduced. In section 4 the refinement of support and confidence framework and its mathematical constraints ie downward closure property is discussed. In section 5 the future direction is given. 2. ASSOCIATIVE CLASSIFICATIONS Associative classification(AC) mining is a promising approach in data mining that utilizes the association rule discovery techniques to construct classification systems. The entire dataset is divided into two part the 70% of data is used as training data and the remaining 30% is used for testing the accuracy of classifier. It is a three-step process shown in figure 1. i. Generate the set of association rules from the training set with certain support and confidence thresholds as candidate rules. ii. Pruning the set of discovered rules to weed out those rules that may introduce over fitting iii. Classification Phase is the step to make a prediction for test data and measure the accuracy of the classifier.