Back

Using Machine Learning along with Data Science algorithms to pre-process and forecast COVID-19 Cases and Deaths

Choudhary, A.

2021-03-17 epidemiology
10.1101/2021.03.15.21253571 medRxiv
Show abstract

The Covid-19 pandemic has taken a major toll on the health and state of our global population. With tough decisions for allocating resources(i.e. vaccines)[1] are being made, forecasting through machine learning has become more important than ever. Moreover, as vaccines are being brought to the public and cases are going down, it is time that we reflect on where the pandemic has taken the most toll:for the purpose of future reform. This research illustrates two different models and algorithms for COVID-19 forecasting: Auto Regressive models and Recurrent Neural Networks(RNNs). The results show the true potential of RNNs to work with sequential and time-series data to forecast future cases and deaths in different states. As the paper utilizes the tanh activation function and multiple LSTM layers, the research will show the importance of machine learning and its ability to help politicians make decisions when it comes to helping states during the pandemic and future reform. The data will also pre-process the time-series data, using rolling statistics and will clean the data for the auto-regressive model and RNN layers. Thus, we show that along with Recurrent Neural Network layers, activation functions also play a crucial role in the accuracy of the forecast.

Matching journals

The top 9 journals account for 50% of the predicted probability mass.

50% of probability mass above

"Similar papers" are the closest papers from that journal in the model's embedding space. They show what the match is built on, but the ranking comes mostly from a classifier over the whole training set, not from these examples alone.