Authors: Ning Chen, Elliot Macy
The goal of this project is to investigate Coronavirus and make related predictions in regard to time series modeling. The data is accessed by public APIs. Machine learning and deep learning methods such as ARMA, ARIMA, SARIMAX, Facebook PROPHET, Recurrent Neural Network and Long short-term memory (LSTM) Networks are implemented and evaluated.
The prediction problem of Coronavirus comes with a significant degree of ambiguity, which is difficult to predict considering the complex circumstances in the real world. While one important task in this project is understanding time series modeling and forecasting related rates for the decision makers to layout some strategies in dealing with covid-19.
The Coronavirus data is accessed by open public APIs without authentication. It provides updated information associated with COVID-19. The public data API provides access to all of the data at a national and state level. The death cases, positive cases and mortality are studied to make further predictions.
ACF and PACF
SARIMAX Model
death prediction
All models are evaluated by RMSE and MAE.
- LSTM Networks serve as the best model for coronavirus prediction with small RMSE and MAE.
- SARIMAX with opitmized hyperparameters by Gridsearch also work well for the prediction.
- Due to limited known condtions, it is difficult to make precise predictions.
- It shows less death and positive cases at weekends and more cases on Monday.
Please review our full analysis in our Jupyter Notebook or our presentation.
For any additional questions, please contact Ning Chen—[email protected], Elliot Macy—[email protected].
Description of the structure of the repository and its contents:
├── README.md <- The top-level README for reviewers of this project
├── project_coronavirus <- Narrative documentation of analysis in Jupyter notebook
├── Project_Presentation.pdf <- PDF version of project presentation
├── data <- Both sourced externally and generated from code
└── images <- Both sourced externally and generated from code