plot | Varsha Ujjinni Vijay Kumar

R Package Review

Mon, 26 Oct 2020 00:00:00 +0000

Overview 🎥

In this blog post, I will be reviewing an R package (How fun is that?), called covid19ViewerShiny, authored by Yuheng Cui, containing different functions, a dataset and a shiny application which depicts the COVID19 pandemic which is currently running rampant throughout the world.

The development version of the package can be found in GITHUB.

The main goal of covid19ViewerShiny package is to a compact and consolidated way for users to visualize and understand the baneful effects of the pandemic. This package has a shiny dashboard that visualizes the condition of the pandemic in all countries of the world. It contains three functions:

launch_app() : This function helps the user launch the shiny dashboard for user interaction and more.
multiSelection() : This function helps the user to create a drop list in the shiny application, which contains country names that the user wishes to view the condition of. This is a function which is for creating the user interaction with the app.
colPlot() : This function is provided to create a column plot within the shiny app which visualizes the cumulative_cases and cumulative deaths of top 5 countries in each category.This function is for removing redundancy within the codes for creation of the plots.

What I think about the package? 💭

From the name, the README file and the pkgdown web page, it was quite clear what this package does, I expected this package to produce a shiny application with just one function, which shows the pandemic rise and the deaths associated within each country, a dataset and also three functions.

And I would say that the package delivered what was spoken of within the documentations and the name. The launch_app() is the main function for this app, it was easy for me to successfully run the function to get a shiny app without any extra work from my end. The files and folders are structured and I could find my way around them with ease.

The shiny app as spoken in the article by the author: covid19 app,contains four tabs within the app.I really like the layout of app, it is not too much information and cramped up. Just enough information for user’s understanding.

The application provides a slider input for the dates reflecting the cases based on the user’s input, a line graph which shows the cases within the countries chosen by the user.

Improvements on the package, shiny app or the documetation? 📝

I would like to start off by saying the package itself as a whole is structured , accomplishes what is expected of it without any errors. Although I did notice that the test for server function colPlot() was not done, tests are very important when making a package, it gives a clear understanding to the developer whether the functions made are effective and working well for any type of input values.

Coming to the shiny app, I really appreciate the efforts the author has taken to make this user effective, interactive app. One thing within the codes of the app is the use of HTML tag <p>, I would like to add that the author does not have to write <p> tag for every line, as the <p> tag does tell us it is a paragraph.

Lastly, the documentation, I was able to navigate myself through this R package thanks to the documentation made. There were three points that I would like to convey:

I found it to be very difficult for understanding how to give the arguments for the function, colPlot(), the function failed for me outside the app and I did try to find out different ways I could use it but in vain , would really appreciate it if that was given more clearly.
After running the command renv::dependencies() in the project, I found that a couple of R package were not mentioned in the DESCRIPTION file Suggests/Imports sections like janitor,usethis,tidyverse and pkgdown. I would recommend using renv::depenedencies on the future packages to know which are the packages you can/should include.
Lastly, the documentation for the shiny app, as not all people who look at the visualizations can take away what the author is trying to convey, maybe a summary of each plot, if the page is completely filled , you could try using tabset for a better understanding of what is.

Takeaways 📖

Other than being amazed by the amount of work put into the package, it is always a learning step when looking into the codes and ideas of peers. Easier way of implementing any code you thought otherwise, so it is always a pleasure to look at the codes which create such fantastic plots. Peer suggestions help develop us into better programmers and help us get better ideas of making any project be seen in a different light. Reviewing this package helped me see the different things I could have added into my own work, to make it be better.

Appendix

As a peer of the author of the reviewed package, Yuheng Cui, it was a pleasure reviewing his work in creating an R package.

The gist of the review is as follows:

Documentation

The package includes all the following forms of documentation:

Repository has a folder called inst/app/ that contains the relevant R code to run the shiny application.
There is an exported function called launch_app() that will find and load the shiny application from inst/app.
Refactor at least one part of the user interface-side logic of your app into R functions that are exported by your R package.
Refactor at least one part of the server-side logic of your app into R functions that are exported by your R package.
At least one unit test for the functions you created to simplify the user interface and server of your app.
All functions in the package are documented with roxygen2 according to guidelines given in lectures.
The DESCRIPTION file includes an informative name and title for your package, as well as a valid author, field and dependencies. There should also be a valid LICENCE file.
A vignette that describes how to launch the app, what the app does, and guidelines for its use.
There is a README.md file that has clear instructions on how to install the package from GitHub, and gives a big picture overview of the package.
There is a pkgdown site that contains the documentation for your package.
The pkgdown site has customized at least one element of the _pkgdown.yaml file.
The pkgdown site is available publicly via deployment from Github Actions.

Functionality

Installation: Installation succeeds as documented.
Functionality: Any functional claims of the software been confirmed.
R CMD CHECK passes without errors or warnings.
Automated tests: Unit tests cover essential functions of the package and a reasonable range of inputs and conditions. All tests pass on the local machine.

Estimated hours spent reviewing:

Should the author(s) deem it appropriate, I agree to be acknowledged as a package reviewer (“rev” role) in the package DESCRIPTION file.

References

Hadley Wickham, Romain François and Lucy D’Agostino McGowan (2020). emo: Easily Insert ‘Emoji’. R package version 0.0.0.9000. https://github.com/hadley/emo

COVID-19 case analysis in India

Thu, 23 Jul 2015 21:13:14 -0500

COVID-19 or coronavirus is the infectious disease caused by the most recently discovered coronavirus. It was first known in December,2019 at Wuhan,China, which then soon turned into a global pandemic as declared by the World Health Organization(WHO) on 11 March,2020.

There are many ways that the virus is currently spreading, but it started out to be spread primarily from person to person through small droplets from the nose or mouth, which are expelled when a person sneezes,coughs or speaks. People can catch the virus if they breathe in these droplets. The most common symptoms of this virus have said to be evolving ever since the pandemic first came into effect, where currently people have been diagnosed with the virus without the initial symptoms like cough,fever,etc.

More Information on COVID-19

Unfortunately, one of the top 5 countries that were affected by the pandemic was India, after United States and Brazil recently with a whooping count of 122437975.This is a relatively high number of people who were confirmed with the covid virus tests, as the population of India is 1210568111, which brings to the conclusion that 10.1140922% of population in India have been tested Positive with the virus.

The DT figure 1 above shows the different states and territories within the country, the number of cases, deaths and recovered cases within each state, this table also shows the total population of each state/territory and the number of hospitals and the number of beds that are available per person within a state/territory for the population, this can be used to predict the efficiency of hospitals within a state/territory to accommodate the people infected. It also shows the difference in population within each state as urban and rural populations and corresponding number of beds within the state. This is a pdf downloadable datatable, all the other data numbers can be viewed by scrolling either right<–>left or up<–>down. You can also highlight the line by just clicking on the row.

Figure 1: Bar plot for visualising the total cases in the states

The highlight of the analysis of this table is to look at the top 5 most affected states/territory within the country:

Maharashtra : 30698938 confirmed cases, with a population of 112374333
Tamil Nadu : 16554118 confirmed cases, with a population of 72147030
Andhra Pradesh : 10102865 confirmed cases, with a population of 49577103
Delhi : 9845140 confirmed cases, with a population of 16787941
Karnataka : 8535942 confirmed cases, with a population of 61095297

This can also be observed in the figure 1 for better visualization.

Table 1: Summary Stats for covid cases in India
	Andhra Pradesh (N=15)	Delhi (N=20)	Karnataka (N=19)	Maharashtra (N=16)	Tamil Nadu (N=32)	Total (N=102)	p value
Date							0.003
- Median	2020-04-20	2020-04-10	2020-04-06	2020-04-13	2020-05-25	2020-04-20
- Range	2020-03-14 - 2020-05-23	2020-03-03 - 2020-05-21	2020-03-09 - 2020-05-24	2020-03-12 - 2020-05-23	2020-03-07 - 2020-06-10	2020-03-03 - 2020-06-10
Cured							< 0.001
- Mean (SD)	408.400 (587.763)	794.600 (1393.547)	136.421 (202.675)	2059.000 (3580.840)	7344.781 (6762.506)	2868.500 (5095.893)
- Range	0.000 - 1763.000	0.000 - 5192.000	0.000 - 608.000	0.000 - 12583.000	0.000 - 18325.000	0.000 - 18325.000
Deaths							< 0.001
- Mean (SD)	21.200 (20.595)	36.050 (47.376)	11.895 (14.008)	355.812 (482.966)	110.938 (100.019)	103.020 (227.632)
- Range	0.000 - 55.000	0.000 - 176.000	0.000 - 42.000	0.000 - 1517.000	0.000 - 307.000	0.000 - 1517.000
Confirmed							< 0.001
- Mean (SD)	976.667 (960.458)	2393.650 (3300.069)	396.316 (515.821)	9363.500 (13743.809)	13984.469 (12167.330)	6542.863 (10470.904)
- Range	1.000 - 2709.000	1.000 - 11088.000	1.000 - 1959.000	11.000 - 44582.000	1.000 - 34914.000	1.000 - 44582.000

Having done the datatable which shows the top 5 states/territories that have been adversely affected by the virus as our examples, we can further analyse the summary statistics for those states(Mean, Range, etc) which helps us understand the rate of infection, this can be produced by using the arsenal package in r for creating descriptive summary statistics.

The table 1 shows the relation of the state/territory with the other variables like death, recovered and ofcourse, the timeline of number of cases recorded.

What does the arsenal table depict?

Well the answer is pretty simple, for every state in the table, the factors are calculated for a p-value which helps in prediction of data, here in our case the case that whether these factors(death,cure,date,etc) affect the number of total confirmed cases within the state and hence the overall country. Looking at the last column within our table 1, we see a p-value of either 0.003 or 0 which is a value very less than the significance level (0.05), which means that the values death, cured, date and confirmed cases are factors of change for the rate of infection within each state.

The means and ranges for each of the factors against the states have also been calculated.

There are certain dates within the cumulative dates that are very significant and stand out for the our analysis.
The means for cured depict the state that had a very high average rate of recovery, which according to the table is Tamil Nadu followed by Maharashtra .
The means for Deaths depict the average deaths within each state and the highest rate of deaths according to the table is in Maharashtra.
The means for Confirmed Cases within each state depict the cases with positive test results with Maharashtra at the top which matches our analysis with the datatable from above.

From the above analysis, we can conclude by saying that the overall situation of the country is not good, with cases increasing everyday and a population so high, precautions are a must. The state with the highest number of recorded cases and deaths with very low recovery rate in accordance to the infection rate is Maharashtra.

I know it is a scary time but we need to take all the necessary precautions for ourselves and our loved ones. Stay Safe!

References

How to Easily Create Descriptive Summary Statistics Tables in R Studio - By Group. Thatdatatho.com. (2020). Retrieved 3 September 2020, from http://thatdatatho.com/2018/08/20/easily-create-descriptive-summary-statistic-tables-r-studio/.

Covid:19 India Data Analysis and Forecasting. Kaggle.com. (2020). Retrieved 3 September 2020, from https://www.kaggle.com/anujgupta872003/covid-19-india-data-analysis-and-forecasting.

Corona Stay Home GIF by INTO ACTION - Find & Share on GIPHY. GIPHY. (2020). Retrieved 3 September 2020, from https://giphy.com/gifs/IntoAction-chart-stay-home-stayhome-Qu1fT51CG14ksIkASL.

Yihui Xie, Alison Presmanes Hill, and Amber Thomas (2017). blogdown: Creating Websites with R Markdown. Chapman and Hall/CRC. ISBN 978-0815363729

Wickham et al., (2019). Welcome to the tidyverse. Journal of Open Source Software, 4(43), 1686, https://doi.org/10.21105/joss.01686

Ethan Heinzen, Jason Sinnwell, Elizabeth Atkinson, Tina Gunderson and Gregory Dougherty (2020). arsenal: An Arsenal of ‘R’ Functions for Large-Scale Statistical Summaries. R package version 3.5.0. https://CRAN.R-project.org/package=arsenal

Yihui Xie, Joe Cheng and Xianying Tan (2020). DT: A Wrapper of the JavaScript Library ‘DataTables’. R package version 0.15. https://CRAN.R-project.org/package=DT

Self Assessment for COVID19 shiny App

Thu, 23 Jul 2015 21:13:14 -0500

Why Shiny?

Shiny helps us turn our analyses into interactive web applications without requiring HTML, CSS, or JavaScript knowledge. Easy to the eyes, makes analyzed visualizations easier to understand and portray to the readers. This can be done using the packages shiny and shinydashboard in R.

Link to the shiny app: COVID19-dash

About the Shiny App?

The above shiny application is a dashboard for the Coronavirus 2019 pandemic throughout the top four countries that is affected by the infection, which are in order as follows :

United States of America with a total of 8,038,543 cases till date.
India with a total of 7,179,006 cases till date.
Brazil with a total of 5,103,408 cases till date.
Russia with a total of 1,326,178 cases till date.

This with the inclusion of all the cases of other countries bring the COVID19 cases to a total of 38,120,172 worldwide according to the website. This application is an inspiration drawn from many a shiny dashboards found online, one of which is from here. The sole purpose of this shiny app is that it might help people who look at the analysis and visualization in understanding the various information related to the pandemic and contribute to the curbing of the virus. There are many small details that the dashboard holds like how to wear a mask, which helps reduce the spread or what symptoms the citizens need to look out for. There are useful links that could help people find more information on the virus. There are 5 windows, out of which the second and the third page show the visualizations of how the spread of the virus is within these 4 countries.

What was done well?

In my opinion,

The formatting of the application is very easy on the eyes, the theme used, the flow of the pages is easy to navigate and justifying of the paragraphs.
The second page is that of leaflet clicks for the change in line plot, which shows the how the virus has changed throughout the period of time for each of the states clicked on the map plot, for each of the four countries.
The third page includes a plotly click for the plotly graph, which shows the total number of cases throughout the four countries, which upon clicking on the specific country gives a DT table with that selected country’s dataset.

Improvements:

I would like to improve on the line graph which shows the number of cases, deaths or the recovered over time.
I could make it to show the information a little better with good titles, geom_texts to the graph, include some markers for the leaflet maps.
I would like to change the DT table to show other details related to the country rather than its own dataset.

Any Additions?

Like I said before too, I would like to add more columns which adds more value to the DT table for the respective click on the plotly plot, I could not think of anything I could add right then but after giving it a lot of thought I think it would be good to add a cases:population ratio for every country during that day.
A plot for how the stringency index increased or decreased over time within each country.
I could have added more features into the application, like a sentiment analysis with each of the countries.

Any Addition that cannot be done yet due to limitation in data or technology?

When I took the dataset from the referenced kaggle websites in the shiny application, these datasets did not contain the latitudes and longitudes for each state within each country, which is why I have added the (lat,long) columns into the datasets by making a separate excel sheets and left joined them to obtain the current result.
If the datasets had a column with the most important dates’ description, I could have added date specific points on the line graph.

References

(2020). Retrieved 14 October 2020, from https://i.pinimg.com/564x/c1/d0/f7/c1d0f79277d914f24d18cd0f1fe0b45a.jpg

Visualising COVID-19 in Peru

Thu, 23 Jul 2015 21:13:14 -0500

On 12 January 2020, the World Health Organization (WHO) confirmed that a novel coronavirus was the cause of a respiratory illness in a cluster of people in Wuhan City, Hubei Province, China, which was reported to the WHO on 31 December 2019. COVID-19 affects different people in different ways. Most infected people will develop mild to moderate illness and recover without hospitalization.

COVID-19 risk is really high in Peru. Its statistics ranks it to be the 5th most affected country in the world after USA, India, Brazil and Russia with 710,067 cases as of September 11th, 2020 by the worldometer website. The timeline of how COVID-19 was first discovered in Peru has been given on Wiki. The virus was reported to have spread to Peru on 6 March 2020, when a 25-year-old man who had traveled to Spain, France, and the Czech Republic tested positive. On 15 March 2020, President Martín Vizcarra announced a country-wide lockdown, closing borders, restricting domestic travel, and forbidding nonessential business operations.

More Information on COVID-19

COVID-19 cases within the country

Figure 1: Count of Covid19 cases(total cases and Deaths)

The plot 1 depicts a line graph for the total cases/deaths on the y-axis and Month/date on the x-axis. Looking at the graph, it can be seen that the cases in Peru did not start until the month of March,2020 and has since exponentially increased consistently till date(Sept 8th,2020) where the count of total cases is 2020-09-08. The graph also depicts the total deaths in the country, where the first death occurred on the 20th March,2020, with respect to the total number of cases, it seems that the total number of deaths are quite less so that is good for the country. Over the months there has been a slight increase in the number of deaths but has not gone more than 30K. There are two dates that has a step wise increase in the deaths(~ 3000 deaths a day) on 24th July,2020 and (~4000 deaths a day) on 14th August,2020 and total number number of deaths currently at 29,976.

Figure 2: Daily new cases

The plot 2 shows a point graph with a line passing through it that helps see how the trends have been since the first record and then the different points indicate the dates on which the cases were recorded. As referred in the introduction, the first case recorded was on March 7th. The number of cases on a daily basis has been changing throughout, the first highest recorded cases was on 1st June with 8805 cases, and notably the very next day there was a huge drop in the cases equal to 0 which is very suspicious and this happens twice before the cases recorded has spiked to a high value of 10143 on 17th August and currently there has been a decrease in the cases all the way to 1598.

COVID-19 case anlysis within each state and it’s population

Figure 3: Highest cases amongst the provinces in Peru

The bar plot 3 with the province names on the y-axis and the number of deaths on the x-axis, with the province LIMA, 10352 number of deaths throughout the months followed by both CALLAO, 1325 and TRUJILLO, 1386 number of deaths in the entire country of Peru.

Figure 4: Difference between the number of deaths by gender amongst the provinces

The bar plot 4 depicts the number of deaths between different provinces in Peru and this has been facet_wrap() by gender. As Lima is the largely populated province in the country of Peru, it has the highest number of cases/deaths recorded. As seen in the graph the number of deaths among males in Lima are higher than the number of deaths among the female gender. Another noticeable feature is that every province under consideration, the males are more prone to have been affected with the virus than the females.

Figure 5: Number of deaths with respect to the age of the person

Lastly, the plot 5 depicts the bar plot of number of deaths occurred within each age group from 1 to 50 facet_wrap() by gender. As seen in 4 the number of deaths among males are comparatively higher than the number of deaths in females. Within which the age group approaching 50 are the most affected age group in Peru, the same can be views among the female gender as well. It seems that children do get affected by the virus but the count is very low ~6 within our dataset.

Stay safe everyone!

References

Ritchie and Hannah (n.d.)
Clark (2020) Wickham et al. (2019) Xie, Hill, and Thomas (2017) Wickham (2016) (2020)

Clark, Martin. 2020. “Peru Covid-19.” Kaggle. https://www.kaggle.com/martinclark/peru-covid19-august-2020.

Ritchie, Research, and data: Hannah. n.d. “Coronavirus Pandemic (Covid-19) – the Data.” Our World in Data. https://ourworldindata.org/coronavirus-data.

Wickham, Hadley. 2016. Ggplot2: Elegant Graphics for Data Analysis. Springer-Verlag New York. https://ggplot2.tidyverse.org.

Wickham, Hadley, Mara Averick, Jennifer Bryan, Winston Chang, Lucy D’Agostino McGowan, Romain François, Garrett Grolemund, et al. 2019. “Welcome to the tidyverse.” Journal of Open Source Software 4 (43): 1686. https://doi.org/10.21105/joss.01686.

Xie, Yihui, Alison Presmanes Hill, and Amber Thomas. 2017. Blogdown: Creating Websites with R Markdown. Boca Raton, Florida: Chapman; Hall/CRC. https://github.com/rstudio/blogdown.

2020. GIPHY. https://media.giphy.com/media/QvMT31LoM66yQmuYbO/giphy.gif.