Abstract
The Lazy learning associative classification (LLAC) is one of the
associative classification methods, in which it delays the processing of
training data until receives a test query, whereas, in eager learning,
the system starts processing the training data before receiving queries.
In this paper, the Lazy Learning Associative Classification with
Weighted kNN (LLAC_WkNN) and Dual Weighted kNN (LLAC_DWkNN) are
proposed. Where LLAC is applied on the dataset, that gives a subset of
rules. Then weighted kNN (WkNN) algorithm is applied on this generated
subset to predict the class label of the unseen test instance. This
yields the improved accuracy of the classifier. The WkNN gives more
weightage to outlier also. This limitation of WkNN is overcome by
applying Dual Distance weighted kNN to LLAC. LLAC_DWkNN checks only k
nearest neighbours, not all the large no of subsets for the subset
evaluation and also gives less weight, which improves the accuracy of
the classifier, further. The comparison results are shown in this paper
of proposed algorithms with the existing associative classification
methods and traditional methods in terms of classification accuracy.