akash87 akash87 - 1 year ago 62
R Question

How to create a different report for each subset of a data frame with R markdown?

I have a dataset that looks like

City Score Count Returns
Dallas 2.9 61 21
Phoenix 2.6 52 14
Milwaukee 1.7 38 7
Chicago 1.2 95 16
Phoenix 5.9 96 16
Dallas 1.9 45 12
Dallas 2.7 75 45
Chicago 2.2 75 10
Milwaukee 2.6 12 2
Milwaukee 4.5 32 0
Dallas 1.9 65 12
Chicago 4.9 95 13
Chicago 5 45 5
Phoenix 5.2 43 5

I would like to build a report using R markdown; however, for each city I need to build a report. The reason for this is that one city cannot see the report for another city. How do I build a report and save a PDF of it for each city?

Each report would need the median
, mean
, and mean
. I know that using
I could simply use

finaldat <- dat %>%
group_by(City) %>%
summarise(Score = median(Score),
Count = mean(Count) ,
Return= mean(Returns))

But the frustration comes from producing a report for each
. Also, this is a subset of the data, not the full data. That is, this report is extensive and is a report of the results, which is systematic, not different for each

Answer Source

It looks like a parameterized report might be what you need. See the link for details, but the basic idea is that you set a parameter in the yaml of your rmarkdown report and use that parameter within the report to customize it (for example, by filtering the data by City in your case). Then in a separate R script, you render the report multiple times, once for each value of City, which you pass as a parameter to the render function. Here's a basic example:

In your Rmarkdown report you would declare the parameter in the yaml. The listed value, Dallas in this case, is just the default value if no other value is input when you render the report:

title: My Document
output: pdf_document
   My_City: Dallas

Then, in the same Rmarkdown document you would have your entire report--whatever calculations depend on City, plus the boilerplate that's the same for any City. You access the parameter with params$My_City. The code below will filter the data frame to the current value of the My_City parameter:

dat %>%        
    filter(City==params$My_City) %>%
    summarise(Score = median(Score),
              Count = mean(Count)  ,
              Return= mean(Returns))

Then, in a separate R script, you would do something like the following to produce a separate report for each City (where I've assumed the Rmarkdown file above is called MyReport.Rmd):

for (i in unique(dat$City)) {
                      params = list(My_City = i),
                      output_file=paste0(i, ".pdf"))

In the code above, I've assumed the dat data frame is in the global environment of this separate R script that renders MyReport.Rmd. However, you could also just provide a vector of city names instead of getting the names from unique(dat$City).