Please use this identifier to cite or link to this item: http://repository.aaup.edu/jspui/handle/123456789/2348
Title: Prediction and Classification Analytics of Obesity Datasets Using a Hybrid Model of Clustering and Neuro-Fuzzy Methods رسالة ماجستير
Authors: Hantoli, Younes Nedal Younes$AAUP$Palestinian
Keywords: artificial neural networks,artificial intelligence,neuro fuzzy models
Issue Date: 2021
Publisher: AAUP
Abstract: Human health is the most important and valuable field in life. Obesity is one of the main causes of many diseases that might cause death such as diabetes, hypertension, and stroke. Artificial Intelligence (AI), which has offered inference tools to support clinical decision-making. The integration of Artificial Intelligence in the field of Healthcare diagnostic is an effective method in a large number of health care applications. The diagnostic procedures for health care problems can be categorized as intelligent data, prediction, and classification tasks. Artificial intelligence techniques can be utilized to predict and classify obesity disease to give appropriate assistance to physicians in decision-making. Several intelligence techniques are being used to diagnose diseases such as a neural network, fuzzy logic, expert systems, etc. In this research, different Artificial intelligence techniques were used to classify and predict the child's obesity. The dataset was collected from 4 cities in Palestine. The Collected data passed through data preprocessing and applying feature extraction that most closely affect the child's obesity. Where the final step is applying the AI methods to recognize the patterns in the dataset. Decision-tree, k-nearest neighbor, support vector machine, logistic regression, neural network, and a hybrid adaptive neuro-fuzzy inference system that combines fuzzy logic and neural networks, were used to recognize the pattern on the dataset and improve the results of the classification of obesity in children with high accuracy. For this neuro-fuzzy hybrid model, the membership functions used are trimf, trapmf, gaussmf, and gauss2mf, where two types of Neuro-fuzzy structures were used; grid partitioning and clustering structure. The total of fuzzy rules was 512 obtaining as an output and the degree of belonging of a child to obesity or not. Based on the obtained results for the applied dataset that represents all cities , the hybrid adaptive Neuro-fuzzy inference system achieved a prediction accuracy of 98.33% using grid partition and using neural networks which achieved an accuracy of 98.40% which are very good results and deserved to be used in a real application to help specialists in making decisions, while the other models got accuracy as follows: Logistic Regression 97.50%, d-tree 92.30%, KNN 93.60%, and SVM 97.10%. Also another results were obtained when the same techniques were applied for each city alone, city 1 results (ANFIS-Grid 99.97% , ANFIS-Cluster 98.58%, Logistic Regression 95.20% and Neural Network 97.40%), the other detailed results are shown in chapter 4.
Description: Master's Degree in Computer Science.
URI: http://repository.aaup.edu/jspui/handle/123456789/2348
Appears in Collections:Master Theses and Ph.D. Dissertations

Files in This Item:
File Description SizeFormat 
يونس حنتولي.pdf2.18 MBAdobe PDFThumbnail
View/Open


Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.

Admin Tools