FluViewR

This package contains various functions to access raw data on CDC FluView through web scraping. Note that some of these data may be available more readily on the CDC Open Data Website. The benefit of scraping is you can more easily obtain basic data. The R package RSocrata is ideal for obtaining data from the open data API.

Available Functions

1. fluviewr_data utlizes the Fluview Interactive API as implemented by the cdcfluview package available on CRAN. The fluviewr_data function essentially does the work of fluview.scrape and fluview.stack outlined below, both of which are being deprecated.

Currently, this function only scrapes nationally aggregate data for all years available (1997 - present) though it will be updated to obtain various geographies in the future.

To use this function, simply call:

df <- fluviewr_data()

The output of this function will contain stacked, cleaned and aggregated clinical and public health data, separated or aggregated by general influenza A vs B as well as the proportion of each for surveillance purposes.

2. fluview.scrape function contains a function to scrape clinical, public health, ILI, or ILI by age group data from HTML tables on the CDC Flu View Website.

fluview.scrape contains two options:

1. loc, which is a character vector of length 1 taking one of the following: “clinical”, “public health”, “ILI”, or “ILI age”- this identifies the corresponding table

2. start, which is a numeric vector of length 1 corresponding to the 4 digit year of the start of the influenza season of interest

A single year can be derived with:

fluview.scrape("clinical", 2017)

This would give you the 2017-2018 influenza season data for clinical reports.

To obtain data for multiple years:

df.clin <-
  rbind(
    fluview.scrape(loc ="clinical", start=2015),
    fluview.scrape(loc ="clinical", start=2016),
    fluview.scrape(loc ="clinical", start=2017),
    fluview.scrape(loc ="clinical", start=2018),
    fluview.scrape(loc ="clinical", start=2019),
    fluview.scrape(loc ="clinical", start=2020)
  )

Alternatively, you can use an apply function or a for loop:

df.clin <- sapply(seq(2015,2020), function(x) fluview.scrape(loc = "clinical", function(x) start=x))

3. fluview.stack takes clinical and public health datasets obtained from fluview.scrape, and cleans/stacks them for analysis.

This function takes two arguments, clin.data and ph.data corresponding to the clinical and public health datasets scraped using the fluview.scrape function

The output is a single dataframe with the total fluA, fluB, total tests, and relevant percentages.

Running the function is simple. Assuming you have a clinical dataset from fluview.scrape called df.clin, and a public health dataset from fluview.scrape called df.ph:

df.combined <- fluview.stack(df.clin, df.ph)

4. fluview.mortplot provides a time series-decomposed anomaly detection plot for Pneumonia and Influenza-associated mortality. This function has no options but uses life data from CDC Flu View. The function will need updated when new influenza seasons are added as the date is hard coded into the URL. The time series is first decomposed using the anomalize package using auto for both frequency and trend options. The decomposition method is twitter and the anomalize method is gesd with 0.05 alpha and 0.2 `max_anomalies. Use our web application to alter these options.

To run the function, do the following:

fluview.mortplot()