Browsing by Author "Sarojini, B"

Now showing 1 - 20 of 37
  • No Thumbnail Available
    Item
    Air Pollution Checker Using Python
    (2022-05) Sofiya Banu, S; Sarojini, B
  • No Thumbnail Available
    Item
    Big Data Framework for Healthcare using Hadoop
    (2010) Sarojini, B
    The technological advancements prevailing today make it possible to sense, collect, transmit, store and analyze the voluminous, heterogeneous medical data. This data could be mined to provide valuable and interesting patterns. However, extracting knowledge from big data poses a number of challenges that need to be addressed. In this paper we propose big data architecture for processing medical data. To handle this large volume of data. Big Data’s Hadoop tool can be used. The Apache Hadoop Framework for processing distributed healthcare database is also discussed.
  • No Thumbnail Available
    Item
    Cancer Detection Using ANFIS and SVM Classification
    (2014) Sarojini, B
    A novel method to enhance the perfomiance of classifiers Adaptive Neuro Fuzzy Inference System (ANFIS), Support Vector Machine (SVM) through feature selection is proposed. The feature selection methods Genetic Algorithm (GA) and Rough Set (RS) are used to select the features. This research work focus on selecting the prominent features to improve the accuracy of the classification algorithms. The experiments are performed on the Breast cancer WBCD and Liver cancer BUPA liver Disorder. The performance of the classification algorithms is estimated in terms of increase in accuracy after feature selection.
  • No Thumbnail Available
    Item
    Comparative Study on the Performance of MMIFS and DMIFS Feature Selection Algorithms on Medical Datasets
    (2015) Sarojini, B
    Background/Objectives: Featxire selection is one of the preprocessing techniques used for removing redundant and irrelevant features. The objective of this research work is to show that small set of relevant features can improve the performance of classification algorithms. Metbods/Statistical Analysis: This paper compares the performance of two feature selection algorithms. Modified Mutual Information based Feature Selection (MMIFS] and Dynamic Mutual Information based Feature Selection (DMIFS) The performance of these feature selection algorithms on the medical datasets is analyzed. The performances of c4.5 classification algorithm before and after feature selection are analyzed. Findings: The comparative study show that the feature selection algorithms have selected prominent features of the medical datasets. The percentage of feature reduction and the improvement in the accuracy of the classification algorithm are used for validation. The result shows an improvement in the accuracy of the classification algorithm. Applications/ Improvements: The reduction in the number of features means diagnosis of the disease with limited number of relevant features. Integrating feature selection techniques and machine learning algorithms will give a better decision making tool which is appreciable in medical domain.
  • No Thumbnail Available
    Item
  • No Thumbnail Available
    Item
    Content Management System For Jewellery Shop Website
    (2015-03) Pavithra, J; Sarojini, B
  • No Thumbnail Available
    Item
    Detection and Discrimination of DDoS Attacks from Flash Crowd Using Entropy Variations
    (2013) Sarojini, B
    Internet is a worldwide network that combines millions local to global scope, private public, academics, business, optical network technologies, government networks. It carries an expandable range of information resources and services which lead to bulk exchange of traffic over the Internet every day. This excessive popularity creates some troubles in the networks. Among them. Flash Crowd and Distributed Denial of Service (DDoS) attacks are the two major events. Web services needs stability and security from these two concerns. There are some methods that can discriminate DDoS attack from flash crowd and trace the sources of the attack in huge volume of network traffic. However, it is difficult to detect the exaet sources of DDoS attacks in network traffic when Flash crowd event is also present. Due to the alikeness of these two anomalies, attacker can easily mimic the malicious flow into legitimate traffic patterns and defence system cannot detect real sourees of attack on time. In this paper, entropy variation, a theoretie parameter, is used to discriminate DDoS attack from Flash Crowd and trace the sources of the DDoS attack. Entropy variation is a theoretic concept which is a measure of changes in concentration of distribution of flows at a router for a given time duration. The proposed strategy is effective and efficiently scalable that has several advantages like memory non intensive, minimum overhead in terms of resources and time, and independent of traffic pattern.
  • No Thumbnail Available
    Item
  • No Thumbnail Available
    Item
  • No Thumbnail Available
    Item
    Dimensionality Reduction using Rough Sets - A Framework
    (2013) Sarojini, B
    The healthcare organizations are marching towards improving the quality of services and the efficacy of patients’ care by adopting data mining technologies. In an effort to turn information into knowledge, they face a lot of challenges when trying to deal with large, diverse, and often-complex healthcare data sources. In the presence of hundreds or thousands of features it is found that only very few of the features predominantly contribute for the medical decision making. The objective of this research work is to prove that a small subset of informative features, selected from a whole set of features, may carry enough information to construct reasonably accurate prognostic models. The focus is on selecting the optimal feature subset, with informative and discriminatory features, required for quality medical diagnosis. The rough set based methods are to be used to find the optimal feature subset of the medical databases that could enhance the Accuracy, the Sensitivity and the Specificity of the classification algorithms.
  • No Thumbnail Available
    Item
    E-Profiling For Employees Recruitment
    (2015-04) Subhashree, S; Sarojini, B
  • No Thumbnail Available
    Item
  • No Thumbnail Available
    Item
    An Efficient Framework for Analyzing Real Time Distributed Health Care Data
    (2015) Sarojini, B
    Real time monitoring o f vital health parameters assure not only early detection and prediction o f the disease but also reduce the health care costs. The technological advancements prevailing today make it possible to sense, collect, transmit, store and analyze the voluminous, heterogeneous big data. This data could be mined to derive valuable and interesting patterns. However, extracting knowledge from big data poses a number o f challenges that are to be addressed. In this paper, we propose an efficient framework for analyzing real time distributed health care data, considering the computation time and space complexities. This framework is designed to discover knowledge from a reduced database with selected features, using a machine learning algorithm. That is. the proposed framework (i) Segments/groups the clinical data records based on similarity (ii) Selects the prominent features that are required for smarter analysis (Hi) Discovers patterns using a machine learning algorithm. The discovered patterns provide valuable information required for real time diagnosis, prognosis and treatment planning.
  • No Thumbnail Available
    Item
    An Empirical Study on the Performance of Integrated Hybrid Prediction Model on the Medical Datasets
    (2011) Sarojini, B
    The medical data are multidimensional and hundreds of independent features in these high dimensional databases need to be considered and analyzed, for valuable decision-making information in medical prediction. Most data mining methods depend on a set of features that define the behavior of the learning algorithm and directly or indirectly influence the complexity of the resulting models. Hence, to improve the efficiency and accuracy of mining task on high dimensional data, the data must be preprocessed. Feature selection is a preprocessing step which aims to reduce the dimensionality of the data by selecting the most informative features that influence the diagnosis of the disease. We propose a feature selection embedded Hybrid Prediction model that combines two different functionalities of data mining; the clustering and the classification. The F-score feature selection method and k-means clustering selects the optimal feature subsets of the medical datasets that enhances the performance of the Support Vector Machine classifier. The performance of the SVM classifier is empirically evaluated on the reduced feature subset of Diabetes. Breast Cancer and Heart disease data sets. The proposed model is validated using four parameters namely the Accuracy of the classifier. Area Under ROC Curve. Sensitivity and Specificity. The results prove that the proposed feature selection embedded hybrid prediction model indeed improve the predictive power of the classifier and reduce false positive and false negative rates. The proposed method achieves a predictive accuracy of 98.9427% for diabetes dataset. 99% for cancer dataset and 100% for heart disease dataset, the highest predictive accuracy for these datasets, compared to other models reported in the literature.
  • No Thumbnail Available
    Item
    Enhancing Medical Prediction using Feature Selection
    (2011) Sarojini, B
    The medical data are multidimensional and hundreds of independent features in these high dimensional databases need to be considered and analyzed, for valuable decision-making information in medical prediction. Most data mining methods depend on a set of features that define the behavior of the learning algorithm and directly or indirectly influence the complexity of the resulting models. Hence, to improve the efficiency and accuracy of mining task on high dimensional data, the data must be preprocessed. Feature selection is a preprocessing step which aims to reduce the dimensionality of the data by selecting the most informative features that influence the diagnosis of the disease. We propose a feature selection embedded Hybrid Prediction model that combines two different functionalities of data mining; the clustering and the classification. The F-score feature selection method and k-means clustering selects the optimal feature subsets of the medical datasets that enhances the performance of the Support Vector Machine classifier. The performance of the SVM classifier is empirically evaluated on the reduced feature subset of Diabetes, Breast Cancer and Heart disease data sets. The proposed model Is validated using four parameters namely the Accuracy of the classifier. Area Under ROC Curve, Sensitivity and Specificity. The results prove that the proposed feature selection embedded hybrid prediction model indeed improve the predictive power of the classifier and reduce false positive and false negative rates. The proposed method achieves a predictive accuracy of 98.9427% for diabetes dataset, 99% for cancer dataset and 100% for heart disease dataset, the highest predictive accuracy for these datasets, compared to other models reported in the literature.
  • No Thumbnail Available
    Item
  • No Thumbnail Available
    Item
    A FUZZY BASED SERVPERF MODEL TO ASCERTAIN RESTAURANT SERVICE
    (2011) Sarojini, B
    The service performance of the restaurant industry can be measured and improved only by understanding the potential gaps in quality. To achieve this, the service quality must be thoroughly studied from both the conceptual viewpoint and service quality measurement. This study explores and identifies the perceived service quality of customers for different restaurant service dimensions through a fuzzy based SERVPERF measurement technique. Also the factors influencing the service quality of restaurant services are identified and suggestions are given for the improvement.
  • No Thumbnail Available
    Item
    Idle Time Management
    (2014-03) Priyadharshini, R; Sarojini, B
  • No Thumbnail Available
    Item
    An integrated approach of feature selection and parameter optimisation of kernel to enhance the performance of support vector machine
    (2015) Sarojini, B
    Mining the big data is a challenging task due to the size of the databases and the comple.xity in maintaining precise and non-redundant data. Classification algorithms need to analyse hundreds of independent features in these high dimensional databases for effective prediction. The performance of classification algorithms could be enhanced data if irrelevant and redundant data are removed. Feature selection algorithms help in identifying prominent features that could enhance the performance of the classifier. Additionally, the classification performance of support vector machine (SVM) could be enhanced by setting appropriate kernel parameters. The kernel parameters of SVM are tuned for each feature subset generated by feature selection and the performance is analysed. The feature subset that enhances the classification performance of SVM is the optimal feature subset of the dataset. Experiments are done on three medical datasets. The empirical results prove that integrating feature selection and optimising the kernel parameters enhance the performance of the SVM classifier. The approach is validated in terms of increase in accuracy and area under receiver operating characteristic (AUC) of the classifier.
  • No Thumbnail Available
    Item
    Lung Nodule Detection Using data Mining Techniques
    (2014) Sarojini, B
    Lung cancer is one of the most common deadliest diseases, it affects both men and women throughout the world. Detection of lung cancer at an early stage can increase the survival rate. For accurate detection it needs to identify efficient features and delete redundancy features among all features. The digital chest film in lung cancer database is divided into normal, benign and malignant. The normal ones indicate the healthy patients (non nodules) and can either benign (non-cancerous) and malignant (cancerous). There are five main phases involved in the proposed work. They are image preprocessing, lung segmentation, nodules detection, feature extraction and image classification. The segmentation algorithm used is OTSU method. Segmentation method used to do accurate segmentation of lung region from the digital lung x-ray images. Classification method is used to detect whether the nodule are either benign or malignant.