|
|
 |
| |
|
|
|
Sequential Pattern Mining using RadixTreeMiner Algorithm and Neural Network-Based Classification |
|
|
|
PP: 1-16 |
|
|
|
doi:10.18576/amis/13S101
|
|
|
|
Author(s) |
|
|
|
K. Poongodi,
A. K. Sheik Manzoor,
|
|
|
|
Abstract |
|
|
| Handling large amount of data arriving from internet-based applications is one of the challenging tasks. Recently, more contributions were made to the data mining algorithms, such as clustering and classification. One of the most commonly-used data mining schemes is Sequential Pattern Mining (SPM). Here, statistically significant and relevant sequential patterns are used for classification purpose, but the complexity grows for the increasing data sizes. This paper introduces a novel approach, namely RadixTreeMiner, for mining sequential patterns from the sequence database, and to classify the data efficiently based on maximal sequential patterns. The proposed RadixTreeMiner algorithm constructs the radix tree from the sequences available in the input database, and then identifies the maximal sequences. Further, the Neural Network (NN) approach is employed in this work for the classification of database based on the maximal sequential patterns. Experimentation of the proposed RadixTreeMiner algorithm uses two standard gene-sequence databases and its performance is evaluated based on the metric Classification Accuracy (CA). From the achieved results, it is evident that the proposed algorithm has better performance with values of 0.9038 and 0.8628 as classification accuracy for both the datasets.
|
|
|
|
|
|
|
|
