![Sample our Mathematics & Statistics journals, sign in here to start your FREE access for 14 days]({"type":"image" , "src" : "/sda/5943/TOC-Subject-Sample-2021-Mathematics-Statistics.jpg"})
Abstract
More than 185 countries get affected in the present SARS-COV-2 (aka COVID-19 or n-Corona virus) epidemic. Experts from different domains are contributing a lot to combat this pandemic effectively. Data analysts have made lots of efforts, IT experts also to forecast the severity of infected cases, the death rate, recovery rate, and other health indicators using various statistical and machine learning models. These forecasting models may motivate and help the policymakers to make the decisions based on these predicted results. Although any prediction on such a pandemic problem is non-monotonic and uncertain, these predictions may help us plan to take the needful actions while dealing with this tragic situation. Lots of symptomatic and asymptotic human behavior are observed while investigating the cases. Also, many countries are making dynamic changes in their policies to combat this pandemic. Therefore, such heterogeneous and dynamic changes in policies and human behavior make the problem of proposing an accurate predicting model more complex. We used machine learning methods to monitor the COVID-19 impact country-wise and developed a FOCOMO model to forecast the severity worldwide till the mid of Oct 2020. We referred to the COVID-19 dataset publicly shared by John Hopkins University. We proposed a purely predictive model to monitor the pandemic (without claiming it as perfect or accurate) into three Phases: i) categorization of countries based on the severity of COVID-19 cases, ii) prediction of daily case reporting, and iii) normalization of flattening the forecasting curve based on deaths and recovery rate. These forecasting models will help monitor and plan the human behaviors and severity of the pandemic country-wise and will provide an opportunity to take some corrective actions for shaping the future.