COVID-19: Analysis, Modelling & Visualization using R

R is a powerful language used widely for data analysis and statistical computing. Inclusion of powerful packages in R has made it more and more powerful with time. One such recent and wonderful package for Covid-19 is ‘covid19.analytics‘.

COVID19.Analytics Package

Introduction

The “covid19.analytics” R package allows users to obtain live worldwide data from the novel Corona Virus Disease originally reported in 2019, Covid-19, as published by the JHU CCSE repository, as well as, provide basic analysis tools and functions to investigate these datasets.

The goal of this package is to make the latest data promptly available to researchers and the scientific community. The package is really handy when getting and visualizing real time COVID data. It provides daily count of cases,recoveries and deaths due to this virus. The predefined functions are even more useful. It has inbuilt GLM models for covid and even genome sequence for viruses. It also includes SIR model which actually shows the importance of social distancing to slowing the spread of the virus.

How to Install this package?

install.packages(“covid19.analytics”)

Loading Package in R

library(covid19.analytics)

Overview of the main functions from the “covid19.analytics” Package:

  • covid19.data: This function is used to read “live” data from reported Covid’19 cases.

mydata = covid19.data(case = “aggregated”, local.data = FALSE, debrief = FALSE)

Arguments:

case:

Argumentdescription
Aggregatedlatest number of cases aggregated by country
Time Series data
ts-confirmedtime series data of confirmed cases
ts-deathstime series data of fatal cases
ts-recoveredtime series data of recovered cases
ts-ALLall time series data combined
Deprecated data formats
ts-dep-confirmedtime series data of confirmed cases as originally reported (deprecated)
ts-dep-deathstime series data of deaths as originally reported (deprecated)
ts-dep-recoveredtime series data of recovered cases as originally reported (deprecated)
 Combined
ALLall of the above
 Time Series data for specific locations
ts-Torontotime series data of confirmed cases for the city of Toronto, ON - Canada
ts-confirmed-UStime series data of confirmed cases for the US detailed per state
ts-deaths-UStime series data of fatal cases for the US detailed per state

local.data : boolean flag to indicate whether the data will be read from the local repo, in case of connectivity issues or data integrity

debrief :  boolean specifying whether information about the read data is going to be displayed in screen

Data Structure:

The Time Series data is organized in a specific manner with a given set of fields or columns, which resembles the following structure:

View(mydata)

Using your own data and/or importing new data sets:

If you have data structured in a data.frame organized as described above, then most of the functions provided by the “covid19.analytics” package for analyzing Time Series data will work with your data. In this way it is possible to add new data sets to the ones that can be loaded using the repositories predefined in this package and extend the analysis capabilities to these new datasets.

names(mydata)

Be sure also to check the compatibility of these datasets using the Data Integrity and Consistency Checks functions.

Analytical & Graphical Indicators:

In addition to the access and retrieval of the data, the package includes some basics functions to estimate totals per regions/country/cities, growth rates and daily changes in the reported number of cases.

tsc = covid19.data(“ts-confirmed”)

  • report.summary: This function is used to summarize the current situation, it will first download the latest data and then summarize the top provinces/cities per case. It results on screen table and static plots (pie and bar plots) with reported information. It can also output the tables into a text file.

report.summary(cases.to.process=”TS”, saveReport = FALSE, graphical.output = TRUE, geo.loc = NULL)

Arguments:

cases.to.processwhich data to process: “TS” –time series–, “AGG” –aggregated– or “ALL” –time series and aggregated–
Nentriesnumber of top cases to display (by default = 10 cases)
geo.locgeographical location to process
graphical.outputflag to deactivate graphical output
saveReportflag to indicate whether the report should be saved in a file
Summary of top-10 Countries
  • tots.per.location : It compute totals per region and plot time series for that specific region/country. It provides static plots: data + models (exp/linear, Poisson, Gamma), mosaic and histograms when more than one location are selected.

tots.per.location(tsc, geo.loc =”India”, confBnd = FALSE, nbr.plts = 1, info=” “)

Arguments:

datadata.frame with *time series* data from covid19
geo.loclist of locations
confBndflag to activate/deactivate drawing of confidence bands base on a moving average window
nbr.pltsparameter to control the number of plots to display per figure
infoadditional info to display in plots’ titles
GLM Plot for Total cases in India
  • growth.rate : It compute changes and growth rates per region and plot time series for that specific region/country. It displays list containing two dataframes: one reporting changes on daily basis and a second one reporting growth rates, for the indicated regions. It also produces static plots: data + models (linear,Poisson,Exp), mosaic and histograms when more than one location are selected.

growth.rate(data0=tsc, geo.loc =”India”, stride=1, info=””)    

Arguments:
datadata.frame with *time series* data from covid19
geo.loclist of locations
confBndflag to activate/deactivate drawing of confidence bands base on a moving average window
nbr.pltsparameter to control the number of plots to display per figure
infoadditional info to display in plots’ titles
A) Changes in Covid-19 cases in India on daily basis
B) Growth Rate in Covid-19 cases in India on daily basis
  • single.trend: It is a function to visualize different indicators for trends in daily changes of cases reported as time series data. It is composed of static plots: total number of cases vs time, daily changes vs total changes in different representations.

single.trend(tsc, confBnd = TRUE, info = “”)

Arguments:
ts.datatime series data
confBndoptional argument to remove the drawing of a confidence band
infoaddtional information to display in plots
Trends in daily changes of cases reported in India

Graphics and Visualization:

  • total.plts: This function is used to plot total number of cases per day for different groups. It produces static and interactive plot

totals.plt(data0 = tsc, geo.loc0 =”India”, one.plt.per.page = FALSE,  log.plt = FALSE, with.totals = FALSE, interactive.fig = TRUE, fileName = NULL)

Arguments:
data0time series dataset to process, default all the possible cases: ‘confirmed’ and ‘deaths’ for all countries/regions
geo.loc0geographical location, country/region or province/state to restrict the analysis to
one.plt.per.pageboolean flag to have one plot per figure
log.pltinclude a log scale plot in the static plot
with.totalsa boolean flag to indicate whether the totals should be displayed with the records for the specific location
interactive.figswith to turn off/on an interactive plot
filenamefile where to save the HTML version of the interactive figure
  • itrends: It is a function to visualize trends in daily changes in time series data interactively.

itrends(ts.data =tsc, geo.loc =”INDIA”, with.totals = FALSE, fileName = NULL)

Arguments:
ts.datatime series dataset to process
geo.locgeographical location, country/region or province/state to restrict the analysis to
with.totalsa boolean flag to indicate whether the global totals should be displayed with the records for the specific location
fileNamefile where to save the HTML version of the interactive figure
Trends in daily changes in cases reported
  • live.map: It generates an interactive map displaying cases around the world.

live.map(data = tsc, select.projctn = TRUE, projctn = “orthographic”, title = “”, no.legend = FALSE, szRef = 0.2, fileName = NULL)

Arguments:
Datadata to be used
select.projctnargument to activate or deactivate the pulldown menu for selecting the type of projection

Projctn
initial type of map-projection to use, possible values are: “equirectangular” | “mercator” | “orthographic” | “natural earth” | “kavrayskiy7” | “miller” | “robinson” | “eckert4” | “azimuthal equal area” | “azimuthal equidistant” | “conic equal area” | “conic conformal” | “conic equidistant” | “gnomonic” | “stereographic” | “mollweide” | “hammer” | “transverse mercator” | “albers usa” | “winkel tripel” | “aitoff” | “sinusoidal”
Titlea string with a title to add to the plot
no.legendparameter to turn off or on the legend on the right with the list of countries
szRefnumerical value to use as reference, to scale up the size of the bubbles in the map, from 0 to 1 (smmaller value –> larger bubbles)
Filenamefile where to save the HTML version of the interactive figure
Live Map displaying cases around

Modelling:

SIR (Suspectible – Infected – Recovered) Model:

An SIR model is an epidemiological model that computes the theoretical number of people infected with a contagious illness in a closed population over time. The name of this class of models derives from the fact that they involve coupled equations relating the number of susceptible people S(t), number of people infected I(t), and number of people who have recovered R(t).

s(t) = S(t)/N,the susceptible fraction of the population,
i(t) = I(t)/N,the infected fraction of the population, and
r(t) = R(t)/N,the recovered fraction of the population.

where N  is the total population 

and at each time  t,  s(t) + i(t) + r(t) = 1

  • generate.SIR.model: It generates a SIR (Susceptible-Infected-Recovered) model based on the actual data of the Covid-19 cases. It provides list containing the fits for the SIR model.

model = generate.SIR.model(data = tsc, geo.loc = “India”,
t0 = NULL, t1 = NULL, deltaT = NULL,
tfinal = 90, fatality.rate = 0.02, tot.population = 130*10^9,
staticPlt = TRUE, interactiveFig = FALSE)

Arguments:
datatime series dataset to consider
geo.loccountry/region to analyze
t0initial period of time for data consideration
t1final period of time for data consideration
deltaTinterval period of time from t0, ie. number of days to consider since t0
tfinaltotal number of days
fatality.raterate of causality, deafault value of 2 percent
tot.populationtotal population of the country/region
staticPltoptional flag to activate/deactive plotting of the data and the SIR model generated
interactiveFigoptional flag to activate/deactive the generation of an interactive plot of the data and the SIR model generated
  • plt.SIR.model: It is a function to plot the results from the SIR model function. It results static and interactive plots.

plt.SIR.model(SIR.model=model, geo.loc = “India”,
interactiveFig = FALSE, fileName = NULL)

Arguments:
SIR.modelmodel resulting from the generate.SIR.model() fn
geo.locoptional string to specify geographical location
interactiveFigoptional flag to activate interactive plot
fileNamefile where to save the HTML version of the interactive figure
Static & Interactive Plot for SIR Model

Make sure you try these functions by yourself. You can just play with the arguments and come up with interesting analysis from the global live data available and thus develop a model which makes good prediction along with awesome graphics.

Have you used covid19.analytics package for your project? or Do you know of any other interesting package? I would love to hear from you! Connect with me in the comments section below or message me on LinkedIn and let’s talk R!

Avatar
I'm an Actuarial Science student having cleared CT1,CT2,CT3,CT5,CS2 & CM2 exams. ➡️Actuarial Trainee- Bharti Axa Life Insurance (Oct 2019 - Present) ➡️Pricing Intern – Bharti Axa Life Insurance (Dec 2018 – Feb 2019) ➡️ Pursuing B.Com from H.R.College of Commerce & Economics ➡️Actuarial Blogger - The Actuarial Club ➡️Research Assistant- Institute of Actuarial & Quantitative Studies ➡️Volunteer at IFOA 400 Club I believe in learning and improving myself everyday. I don't work hard to achieve things instead I do it because I love to! I aim to become an actuarial fellow and use my analytical skills for the greater good someday.

Leave A Reply

Your email address will not be published. Required fields are marked *

This site uses Akismet to reduce spam. Learn how your comment data is processed.

Subscribe us for more