Exact flu spreading prediction can be helpfull in prevention and intervention in case of influenza outbreak. In this work we will describe algorithms of data mining on the case of flu spreading prediction. We will present two models: LASSO and random forest. For each model we will observe a regular forecast and a rolling forecast. For prediction we will use data from social network Twitter, data gathered in Google search queries, history of total number of influence patients and weather data, including temperature and humidity. We will find optimal model for flu spreading prediction and show which set of data gives the best results.
|
