March 20, 2014

Don't you miss popular packages?

This post is continued from the previous post. In the post, I found unique packages which other people had not download. Today, I will check the packages which are popular (best 50) but I missed.

Here's the result.

# All my installed packages
res <- data.frame(name = row.names(installed.packages()), installed.packages(), 
    stringsAsFactors = FALSE)
res <- subset(res, !(Priority %in% c("base")))

# package ranking 2013 from Rstudio log
packageRanking2013 <- read.csv("http://dl.dropboxusercontent.com/u/956851/RStudio_CRAN_data.csv", 
    as.is = TRUE, encoding = "UTF-8")

# result
range_rank <- 50
res_diff <- setdiff(as.character(packageRanking2013$package)[seq_len(range_rank)], 
    res$name)
cat("The number of the packages you missed are", length(res_diff), "(", round(length(res_diff) * 
    100/range_rank, 1), "%)")
## The number of the packages you missed are 13 ( 26 %)

packageRanking2013_missed <- subset(packageRanking2013, package %in% res_diff)
print(packageRanking2013_missed)
##     package count
## 16  mvtnorm  1752
## 21    Hmisc  1604
## 25       sp  1477
## 32      rgl  1224
## 34 multcomp  1193
## 37 sandwich  1120
## 41   lmtest  1036
## 43    RODBC   978
## 45 quadprog   911
## 46  reshape   888
## 47    leaps   875
## 49     coda   856
## 50    abind   839

So what?

The numbers on the lefe side of each packages are ranking. Hmm… There seem to be three types in the above list.

  1. No need: Hmisc, sp, rgl, multcomp, RODBC, reshape
  2. Certain installed: mvtnorm,
  3. Not know to exist: sandwich, lmtest, quadprog, leaps, coda, abind

As for type 1, I rarely analyse data related to the social sciences, so I know Hmisc is a usuful package but I have not installed it. Next, mvtnorm, which is classified as type 2, I'm certain that I have installed it already. I have used this package. Maybe I misunderstood the environment as another environment. The last type includes the packages that I did not know about. However, they are popular among other R users.

Now, by checking package popularity, I have found some interesting points. How about you?

No comments:

Post a Comment