Data Mining and Knowledge Engineering

L. Arockiam ¹, V. Arul Kumar ¹

Affiliations
1 Department of Computer Science, St. Joseph's College, Tiruchirappalli-620002, IN

Abstract

Dimensionality Reduction (DR) is an important technique which is used to reduce the dimensionality of features present in the datasets. This technique is used in various fields such as Data Mining, Machine Learning, Pattern Recognition, Image Retrieval, Text mining etc. In the data mining filed, DR is an important preprocessing technique. Linear Discriminant Analysis (LDA) is a popular DR technique. Traditional LDA technique faces a Small Sample Size (SSS) problem. The SSS problem occurs when the number of samples is less than the dimensionality of the samples. A Lot of feature selection algorithms are proposed in the earlier days, but still the problem persists. Hence, a new feature selection algorithm is proposed in this paper to overcome the SSS problem.

Keywords

Feature Selection, Filter Approach, Fisher Criterion, Feature Selection Algorithm.

Full Text

     

V. Arul Kumar ¹, L. Arockiam ¹

Affiliations
1 Department of Computer Science, St. Joseph's College, Tiruchirappalli-620002, IN

Abstract

Feature selection has become an emerging research area in the field of pattern recognition and machine learning. It is one of the most important processes in Knowledge Discovery. The data set contains irrelevant, redundant and noisy data, which can be preprocessed using feature selection technique. Through feature selection technique the relevant features are identified for the mining process. Feature selection is one of the factors to classify the data without any misclassification and address the performance of the model. In this study, an attempt is made to review the different feature selection techniques in machine learning scheme.

Keywords

Feature Selection, Supervised Learning, Unsupervised Learning, Semi Supervised Learning.

Full Text

     

L. Arockiam ¹, S. Charles ¹, C. Lalitha ¹, A. Sujanitha ¹, V. Arul Kumar ¹

Affiliations
1 Department of Computer Science, St. Joseph’s College, Tiruchirappalli-620002, IN

Abstract

Juvenile Plantar Dermatosis (JPD) known as sweaty sock syndrome. It affects the skin, which becomes scaly and red on the soles of the feet of the children and the young teenagers. A new framework is developed to provide an offline solution for JPD using descriptive and predictive modeling. The symptoms and stages of JPD are classified using clustering and decision tree technique. In clustering technique, the patients’ objects are categorized using symptoms. The decision tree method has been employed for finding the decision rule based on symptoms, which is used for recommendation. The recommendations are useful to the patients those who are affected by the JPD disease.

Keywords

K-Means, Clustering, JPD, Decision Tree.

Full Text