Forecasting the COVID-19 End in India Using Machine Learning and Population Density Clustering

In the literature, the traditional machine learning models are vastly explored in prognostic modelling for COVID-19. The goal of predictions is to make the citizens aware of potential consequences that can be caused by COVID-19. Nonetheless, conventional prediction models are unable to exploit several factors such as demographics and population density. This work addresses how these continuous and unprecedented factors lead us to develop complex models rather than just relying on them. In addition to that based on the obtained data till 3rd September 2020, the state-wise COVID-19 end prediction is also portrayed for India. Further, we have studied the relationship between the population densities and the number of infected cases. We have observed a strong correlation in Central zone, Eastern Zone (EZ) and North Eastern zone (NE-Z) and low correlation in Western Zone (WZ), Southern Zone (SZ), Northern Zone (NZ) and all the states of India. The popular yet powerful extreme learning machine (ELM) model is applied as a forecasting model. The estimation capacity of ELM is compared with the large margin distribution-machine based regression and the twin support vector regression based on R2 and root mean square error. Experimental outcomes prove the proficiency of the ELM model.