covid-19-prediction

Prediction of the evolution of a pandemic using a simplified Susceptible-Infectious-Recovered model

Run it

./covid-analysis.py --help prints out the help:

usage: covid-analysis.py [-h] [--url URL] [--name-dataset NAME_DATASET]
                         [--regions-field REGIONS_FIELD]
                         [--date-field DATE_FIELD]
                         [--variable-name VARIABLE_NAME]
                         [--start-days-ago START_DAYS_AGO]
                         [--days-analyzed DAYS_ANALYZED]
                         [--days-forecasted DAYS_FORECASTED]
                         [--smoothing SMOOTHING] [--json-name JSON_NAME]
                         [--debug]

optional arguments:
  -h, --help            show this help message and exit
  --url URL             URL or filename. Must be a CSV.
  --name-dataset NAME_DATASET
                        Prefix of files/directories created.
  --regions-field REGIONS_FIELD
                        In the CSV, the field with regions.
  --date-field DATE_FIELD
                        In the CSV, the field with the dates.
  --variable-name VARIABLE_NAME
                        In the CSV, the field with the variable we want to predict.
  --start-days-ago START_DAYS_AGO
                        Run forecasts also in the past, starting N days ago.
  --days-analyzed DAYS_ANALYZED
                        Number of days used for the fit
  --days-forecasted DAYS_FORECASTED
                        How many days in the future.
  --smoothing SMOOTHING
                        How many days used for smoothing (poss >= DAYS_ANALYZED).
  --json-name JSON_NAME
                        Name of the JSON file with all data.
  --debug

Example: Italian Regions

Data from Italian Civil Protection:

./covid-analysis.py --smoothing 14 --days-analyzed 14 --start-days-ago 7 --url https://github.com/pcm-dpc/COVID-19/raw/master/dati-regioni/dpc-covid19-ita-regioni.csv --name-dataset protciv --regions-field denominazione_regione --date-field data --variable-name totale_ospedalizzati --json-name protezione_civile_all_data.json

Example: Whole world

Data from Johns Hopkins University Center for Systems Science and Engineering:

./join_time_series.py (convert to a normalized format)

./covid-analysis.py --smoothing 14 --days-analyzed 14 --start-days-ago 7 --days-forecasted 120 --url time_series_adjusted.csv --name-dataset CSSEGI --regions-field "country" --date-field date --variable-name infectious --json-name CSSEGI_all_data.json

Origin of the model

See: COVID-19: Evidence that Italy has done it right (English) or Stima veloce dell’andamento del numero di ricoverati con sintomi COVID-19 in Italia (Italian).