1 Objective: COVID 19 Forecasting
As the covid-19 pandemic is changing the world as we know it, a fundamental problem facing the medical and research community is short/medium term forecasting of covid-19 data (cases, hospitalizations, intensive care bed and deaths). In this project you will build a covid-19 data forecasting engine using neural networks and python.
Guidelines:
- You can choose whatever country you want. You can also choose more than one country or the world at large.
- You can choose whatever output(s) you want to predict: cases, hospitalizations, intensive care occupancy, etc.
- You can choose whatever sources for input you want. These may include past data of cases, mobility data, population density data, healthcare data (e.g. vitamin D intake, BCG vaccine), social network data (e.g. geolocated sentences that indicate symptoms), etc. Be creative on what source you use. Some potential sources can be found here:
o https://coronavirus.data.gov.uk
o https://ourworldindata.org/coronavirus-testing
o http://www.salute.gov.it/portale/nuovocoronavirus/dettaglioContenutiNu
ovoCoronavirus.jsp?lingua=english&id=5367&area=nuovoCoronavirus
&menu=vuoto
o https://datastudio.google.com/reporting/3ffd36c3-0272-4510-a140- 39e288a9f15c/page/U5lCB
o https://www.worldpop.org/events/china
o https://www.google.com/covid19/mobility/
β’ Your report should be done in HTML and put available online (this could be put in your home directory public_html of your account in the departmental servers). It should include the following sections:
- Introduction: What is the problem you are trying to solve and the general
framework to solve it. 1 to 2 pages
- Data sources: explanation of what these are, how they were collected and
how you pre-processed the data
- Machine learning framework and code. This should use neural networks.
- Results. You should present all your results results you may have but you
need to present a table with the accuracy of your results for several days in advance as the one in the following:
- Days in advance
- 1
- 2
- … β’n
β’ Accuracy
β’ 90% β’ 80% β’ 70% β’…
β’ Conclusions
In the submission form you just put the URL of the report.
2 Evaluation
Your project will be assessed on the basis of the following:
70% on the quality of your model and code. How you convert the inputs, activation functions, loss function, etc.
20% on your chosen sources and how they were pre-processed
10% on presentation of your report/page.
