Prediction for Pulmonary Disease Based on Diagnostic Reciepes and Classification

In this research work we have developed a strategy in which the various parameters that influence the occurrence of pulmonary disease have been gathered from survey of doctors who specialize in diagnoses of pulmonary disease and diagnostic recipes involving if the else rules were built and given labels, which were used as target for machine learning algorithms [Logistic , SVM, RBF, Naïve Bayes ] for identification of input dataset of symptoms of subjects . Multiple designs of these classifiers were implemented and best possible machine algorithm was identified for implementing the complete methodology. Results shows that there was no absolute answer for the design and selection of best possible machine algorithm as evident from the results based on multiple statistical tests, therefore , distance from ideal values of statistical test to find best classifier with most optimized parameters was calculated and the classifier which had closest to these ideal values was found and declared the best classifier for identification of pulmonary diseases presence or absence .as per results naïve bayes classifier is performing best which is evident from the statistical test scores .