No abstract

This paper focuses on improving decision tree induction algorithms when a kind of tie appears during the rule generation procedure for specific training datasets. The tie occurs when there are equal proportions of the target class outcome in the leaf node s records that leads to a situation where majority voting cannot be applied. To solve the above mentioned exception, we propose to base the prediction of the result on the naive Bayes (NB) estimate, k-nearest neighbour (k-NN) and association rule mining (ARM). The other features used for splitting the parent nodes are also taken into consideration.

This paper proposes a new classification algorithm named the Attribute Value Dependant classifier (AVD). This AVD classifier mainly focuses on identifying the relationship between the attribute values in the training dataset and the class label values. The ‘Attribute Value Dependant’ is the value which determines the extent to which each attribute value has its impact in deciding the class label is identified; and the training model is built based on these values. The attribute values in the test dataset are compared against the ‘Attribute Value Dependant’ values and based on this comparison the test dataset is classified. Since individual attribute value dependency on the class label is found, the irrelevant and inconsistent data in the training dataset are ignored during the classification and consequently the classification accuracy is improved. The performance of the AVD classifier has been compared with seven traditional classifiers and has been proved to produce better results.