Data Analytics with Efficient Mining of Frequent Patterns on Decision Making
Publication Date : 27/10/2020
Feature and variable selection have become the most focused areas of application in research for which datasets with many number of variables are available. Feature selection in neural network can select features which are essential and discard unwanted and indifferent features. Such a method may pick up some useful but dependent features, all of which may not be needed. The proposed scheme, named as Feature Selection Multi-layer Perception (FSMLP), uses controlled redundancy for selecting the features both for classification and function approximation or prediction type problems. Several data sets like synthetic data set are used to demonstrate the effectiveness of the algorithms. In order to control the redundancy a measure of linear dependency is considered. Nonlinear measures of reliance, such as correlative information are used since they are straightforward. These methods can report for feasible nonlinear fine interactions between tools, as well as that between attributes and the problem being cracked. They can also manage the level of idleness among the selected attributes.
No. of Downloads :