Throughout this project, some notebooks were developed in order to study, through different parameters, the epidemiological situation of the coronavirus in Italy, the European country most affected by the virus.
The dataset used is part of a study carried out by Imperial College London.
Throughout the development of the work, this dataset has undergone some changes that can be consulted in the folder Dataset.
In this project, different Jupyter notebooks were developed:
🗂️ Mapping covid-19 number of cases in Italy (.ipynb)
🗂️ Mapping covid-19 number of deaths in Italy (.ipynb)
🗂️ Statistical analysis of covid-19 data in different regions (.ipynb)
🗂️ Forecasting models (.ipynb)
🗂️ Covid19 Animation - Number of cases per region (.ipynb)
On this map it is possible to see the distribution of the number of cases across the different regions of Italy. If you want to visualize the distribution of the number of deaths in the different Italian regions, just consult this image
Throughout this area of study, numerous graphics have been made that can be consulted in this nootbook
Through these examples, it is possible to compare the most affected regions with each other, through numerous criteria.
With the equivalence of values in numerous criteria, between Piemonte and Emilia-Romagna, there was a need to study the different areas and understand where these two regions stood out along the covid19.
As Lombardia and Piemonte were in the first and second place in the Top 5 of regions with the highest number of cases, we sought to understand the differences between these two regions, relative to the place of recovery/hospitalization.
When using a forecasting model, it is possible, through a curve adjustment, to find the ideal combination of parameters, capable of minimizing errors. Throughout the development of this Jupyter notebook, numerous forecasting models have been developed. In the figure below it is possible to see the number of deaths in Molise (black dots) and the three models (blue - logistic, light blue - exponential and dark blue-linear) that try to fit the Molise curve. With the observation of the model it is possible to verify that the exponential model is the most adequate.
Accordingly, it was also possible to find the ideal parameters for the four models.
To finish the work, a gif was developed capable of demonstrating the daily growth in the number of cases of covid 19, in the top 10 of regions most affected by italy.
Francisca Nogueira A81590 (Biomedical Engineering student in the Master of Medical Informatics, at the University of Minho)
24/05/2020