计算 ggplot2 中堆叠条的累积和

Question

Assume an R data frame ( testData ) that contains three columns (named DATE , FREQ_RECORDS , and CRITERION ) and paired data (column CRITERION contains the values "positive" or "negative").

testData = structure(list(DATE = structure(c(18140, 18140, 18170, 18170, 18201, 18201), class = "Date"), FREQ_RECORDS = c(57L, 120L, 302L, 64L, 40L, 20L), CRITERION = structure(c(1L, 2L, 1L, 2L, 1L, 2L), .Label = c("positive", "negative"), class = "factor")), row.names = c(395L, 756L, 396L, 757L, 397L, 758L), class = "data.frame")

I would like to visualize the data via ggplot2 as dodged bars that are cumulative within (but not across) the pairing factor (ie, the final bars should have a height of 57+302+40=399 for "positive" and 120+64+20=204 for "negative").

I incorrectly believed that the following code would produce such a plot:

ggplot(data=testData, aes(x=DATE, y=cumsum(testData[,"FREQ_RECORDS"]), fill=CRITERION), width=1) + 
    geom_bar(stat="identity", position="dodge", alpha=0.5) + 
    theme_minimal()

What is incorrect about the above code in order to obtain the desired result and how would I need to correct it? Note: I believe it is an issue of how the cumulative sum is calculated (ie, cumsum(testData[,"FREQ_RECORDS"] ), but am uncertain about the details.

Answer 1

When you do cumsum(testData[,"FREQ_RECORDS"]), it is applied onto all of FREQ_RECORDS. The grouping by fill=.. will separately your x and y values accordingly and plot..

So maybe try this, unfortunately you cannot plot it on the fly (I think):

df<-testData %>% 
group_by(CRITERION) %>%
mutate(CUMFREQ=cumsum(FREQ_RECORDS))

ggplot(data=df, aes(x=DATE, y=CUMFREQ, fill=CRITERION), width=1) + 
    geom_bar(stat="identity", position="dodge", alpha=0.5) + 
    theme_minimal()