Heart Disease Classification By Genetic Algorithm


Abstract
Classification is predicting of a correct output for related inputs. This is done by using many techniques, some of them depends on a training process that uses set of features and its related output. In the training process, the algorithm finds the relationships between the features and its related output. In this work, classification heart disease is done using genetic algorithms. Genetic algorithm is used to get best successful system of classification. The chromosome consists of the states using for classification, this means every gene acts state from database. Database is clustered to chromosome to find fitness of chromosome(accuracy of classification).
Key words : genetic algorithms, clustering, patterns recognition.


Introduction
The classification of data is based on the set of data features used. Therefore, feature selection and extraction are main in optimizing performance, and strongly affect classifier design. Defining suitable features often requires interaction with experts in the application area. Genetic algorithms are good candidates for this task since GAs are most useful in multiclass.
Genetic algorithm using to mine a classification rules in large datasets proposed by Vivekanandan and Nedunchezhi(2010).

Building a rule based classification model for these huge data sets using Genetic Algorithm becomes an extremely complex process. They build a model incrementally. An incremental Genetic Algorithm was evolved small components by evolution of the data set which reduce the cost of learning and making it suitable for large data set to build the rule based classification model in a fine granular method .
Parallel Genetic Algorithm depended on Clustering have been proposed by Kanungo et al.,(2007) was used to classify the background and objects. They used a method uses the histogram of the original image where Parallel Genetic Algorithm depended on clustering notion was used the discrete nature of the histogram distribution to determine the optimal threshold.
Fukunaga (1990) divided the space into the parts of classes and defined a problem of evaluating density functions in a high-dimensional space as pattern recognition.