Time series forecasting for number of hospitalizations caused due to COVID-19 in the United Kingdom

Abstract

Pandemics and epidemics have plagued humanity throughout history. The modern world faced one such devastating disease in 2019 called Coronavirus. As the world is still trying to recover from Coronavirus, an epidemic that might not see its end anytime soon. This study focuses on analyzing and predicting the future  hospital admissions that arise due to Covid-19 .For this study, the choice of country is the United Kingdom. The data has been procured from reliable internet sources to carry out all necessary experiments. The data set contains daily hospitalisations due to Covid-19 in the United Kingdom. To carry out the time-series forecasting, the predictive model is built using a special Recurrent Neural Network, also known as Long Short-Term Memory (LSTM) . The final model was built using a Stacked LSTM to predict the number of hospitalisations due to Covid-19 that may arise in the United Kingdom for the next twenty days from the last day of the dataset used. The results of this study show a clear indication that a spike in the number of hospitalisations may arise in the upcoming days.