Missing Value Imputation using Hybrid K-Means and Association Rules

Association rules is an important and well researched database mining function for discovering interesting relationship between variables in large database. The data mining architecture works on facts and figures which are used for any type of decision making. To perform any analysis and decision making, these facts must be complete so that the analyst can make a strategy for decision making. In fact the most important problem in knowledge discovery is the missing values of the attributes of the dataset. If such imperfections are there in database, it is cleaned during pre-processing and is prepared in order to be functional. Considering, the importance of handling missing instances in data mining and knowledge discovery, we proposed a hybrid algorithm for using a combination of association rules mining and k-nearest neighbour approach. We performed a detail experimental result on UCI datasets to check the effectiveness of our technique. For this, we have discretize the data using K-means technique and to generate rules in large volume of data using apriori association rule mining. The measurement metrics such as confidence, support, lift & coverage are then used to evaluate the discovered rules.