Artificial Intelligent Systems and Machine Learning

R. Sridevi ¹, E. Ramaraj ²

Affiliations
1 Manonmaniam Sundaranar University, Tirunelveli, IN
2 Department of Computer Science and Engineering, Alagappa University, Karaikudi, IN

Abstract

Association rule is a method for discovering interesting relationships between the items in large databases. For analyzing the students’ behaviour, the systems accumulate a large volume of valuable information. Since the student database includes more number of attributes it is difficult for processing. The goal of the Multidimensional Quantitative Rule Generation is to generate association rules that satisfy the minimum confidence threshold. But in some cases measuring confidence alone is not sufficient for decision making. Therefore, the Confidence measure for continuous data can be derived that agrees with the standard confidence measure while applying to binary data also. Besides we have taken one more add-on factor `Lift Ratio' which is to validate the generated Association rules that are strong enough to infer useful information. This proposed approach aims to put together the above points to generate an efficient algorithm to offer useful rules in an effective manner.

Keywords

Data Mining, Association Rules, Multidimensional, Confidence, Lift Ratio.

Full Text

     

P. Radha ¹, E. Ramaraj ², S. Purushothaman ³

Affiliations
1 Department of Computer Science and Engineering, Alagappa University, Karaikudi, IN
2 Computer Center, Alagappa University, Karaikudi, IN
3 Udaya School of Engineering, 629204, IN

Abstract

This paper discusses the preprocessing and classification of depression data using back propagation algorithm (BPA). In general, input vectors will not be orthogonal to each other. The proposed method of preprocessing the input vector makes possible BPA learn the input vectors. The classification performance of BPA have been shown for a minimum 80%.

Keywords

Depression Data, Back Propagation Algorithm, Polynomial Discriminant Vector (PDV).

Full Text