Knowledge Data Analysis on Migraine Headaches by Using Optimal Classifiers

Abstract

Machine learning (ML) is largely used to develop automatic predictors in migraine classification but automatic predictors for medication overuse (MO) in migraine are still in their infancy. This research work finds Naïve Bayes and Naïve Bayes Updateable classifier of statistical learning are producing an efficient results compare with functional learning models. In Statistical learning Naïve Bayes and Naïve Bayes Updateable algorithms are having same as well as highest efficient outcome which is 93.50% of accuracy; In Functional learning, Quadratic Discriminant Analysis is having 93% of accuracy which is highest efficient outcome compare with other functional models. In Statistical learning Naïve Bayes and Naïve Bayes Updateable algorithms are having same as well as highest efficient outcome which is 0.93 of PPV; In Functional learning, Quadratic Discriminant Analysis is having highest PPV which is 0.93 of PPV. In Statistical learning Naïve Bayes and Naïve Bayes Updateable algorithms are having same as well as highest efficient outcome which is 0.94 of TPR; In Functional learning, Quadratic Discriminant Analysis is having highest TPR which is 0.93 of TPR compare with other functional learning. In Statistical learning Naïve Bayes and Naïve Bayes Updateable algorithms are having same as well as highest efficient outcome which is 0.99 of AUCROC; In Functional learning, Quadratic Discriminant Analysis and Multi-Layer Perceptron are having highest AUCROC and same AUCROC value which is 0.97 of AUCROC compare with other functional learning. In Statistical learning Naïve Bayes and Naïve Bayes Updateable algorithms are having same as well as highest efficient outcome which is 0.97 of AUCPRC; In Functional learning, Quadratic Discriminant Analysis is having highest AUCPRC value which is 0.93 of AUCPRC Compare with other functional learning. In Statistical learning Naïve Bayes and Naïve Bayes Updateable algorithms are having same as well as highest efficient outcome which is 0.93 of F1-Score;. In Functional learning, Quadratic Discriminant Analysis is having highest F1-Score value which is 0.93 of F1-Score value Compare with other functional learning. In Statistical learning Naïve Bayes and Naïve Bayes Updateable algorithms are having same as well as highest efficient outcome which is 0.89 of k value; In Functional learning, Quadratic Discriminant Analysis is having highest k value which is 0.88 of k value Compare with other functional learning. In Statistical learning, Naïve Bayes and Naïve Bayes Updateable algorithms are having same as well as highest efficient outcome which is 0.90 of phi coefficient value; In Functional learning, Quadratic Discriminant Analysis is having highest phi coefficient value which is 0.88 of phi coefficient value Compare with other functional learning. This work finds that statistical learning models produce better outcome with lowest errors compare with functional model