Exploring Machine Learning Utilization on Influenza Pandemic Dataset

Abstract

There must be an exact system for monitoring the influenza outbreaks to have an optimum solution for the recovery of infected people’s health. For reducing the spread of future outbreaks of influenza virus, forecasting plays an important role. Influenza a is type of disease which is transferred to human beings through pigs, found in animals. It became pandemic in Spain, approximately, 1/3rd of human population died and 1/4th of pig population. Again in 2009, influenza "A" caused millions of deaths, and spread like a pandemic rapidly. Variety of researches inspected data obtained from World Health Organization and local hospitals at country level. This research work is based on mathematical biology using data science techniques in the domain of machine learning. This research suggests a modeling scheme for influenza pandemic predictions, its different classifications and types such as H1N1, B-Victoria etc. via machine learning prediction and regression as well as classification algorithms such as Logistic Regression (LR), Support Vector Machines (SVM) using Linear, Polynomial and RBF kernels; Naïve Bayes (NB) and Random Forest (RF) method for the prediction of influenza disease and its outbreak, the influenza kind became pandemic with the infected populated area. After using various kernels in SVM algorithm, it is observed that Polynomial and Linear kernels have approximately the same accuracy scores, while RBF kernel was not best-fitted for the considered influenza datasets. As far as the overall performance is concerned, at average, RF has the highest accuracy score as 74% while the LR had also the better average score as 72% after RF. After applying the considered ML algorithms, Random Forest algorithm performed in well-effective manner and comparatively it was analyzed as the best-fitted algorithm for the considered datasets.