R-Guru Resource Hub

Home
Compare with SAS

C. Compare with SAS

This section discusses the similarities and differences between SAS and R. While most everything from SAS can be replicated in R, there is a steep learning curve since R concepts and process flow are more object oriented. R has meanings for special characters such as [], {} and () for example. In addition, most of R syntax consists of functions which are similar to SAS functions. So, knowing how to call SAS functions will help to understand, write and execute R functions. Few SAS and similar R terms are listed below.

SAS R
data set data frame
variable vector
data steps, procedures & functions R packages and functions (tidyverse)
proc sql dplyr
macro programs R user defined functions
ODS R Markdown

R has SQL query type functions that enable filter records, subset variables and summary processing. Finally, R processes functions as a collection of independent sequence of steps. So, for example, variables can be created one-by-one with left variable assignments or combined when wrapped in a R function.

With so many SAS programmers learning R, it makes sense that an R SAS package was created. With the SASSY package, SAS programmers can have a more seamless transition between R and SAS. With the SASSY package, SAS programmers can almost replicate reviewing logs, reviewing datasets, and program data steps, formats and reports. There are R packages and functions to replicate Proc Freq, Proc Means and Proc Report. It is important to realize that although SAS has tools to integrate with R and R has packages to replicate SAS programming, the objective of learning R is to consider it a standalone complete toolbox in its own right that can be used entirely independent of SAS.

From SAS to R Results Workshop: Derivations, Queries and Joins (R-Guru Cheat Sheet)

Create libnames [haven]
Import SAS Datasets into Data Frames [haven]
Pipe R Functions [%>%]
Data Management Operations [tidyverse, stringr]
Derive Variables [dplyr]
Query Data Frames [dplyr]
Join Data Frames [dplyr]
Transpose Data Frames [tidyverse]
Create SAS Data sets/Transport Files [xportr]

From SAS to R Results Workshop: Summary Tables (R-Guru Cheat Sheet)

Create libnames [haven]
Import SAS Datasets into Data Frames [haven]
Pipe R Functions [%>%]
Summarize Data (Sex and Age) from ADaMs [dplyr]
Perform T-Test Analysis (Sex by Treatment) from ADaMs
Create Summary Tables as RTF files [gt(), gtsummary()]
Create Plots [ggplot2]
Create SAS Data sets/Transport Files [xportr]

Similar to SAS, R functions resemble many features of the Data Step.

Compare and Contrast SAS Procedures and R Functions

proc print data=d1; var _all_; run; d1 displays all records, head

(d1) displays first 5 records, tail(d1) displays the last few records

To subset, first create R object based on subset and selected variables and then display all records

proc freq data=d1; tables sex*race; run;

table(d1$sex, d1$race)

prop.table()

proc univariate;

summary(d1)

proc sort data=dm2; by sex race; run;

dm2[order(dm1$sex, dm1$race)]

proc format;

In Vectors: sex_code <- c(‘M’, ‘M’, ‘M’, ‘F’, ‘M’) # 1. data values in simple vector to store data values

sex_decode <- c(‘M’=’Male’, ‘F’=’Female’) # 2. named vector data = ‘label’ for values similar to proc format

sex <- sex_decode[sex_code] # 3. converts values to labels sex_code vector is subset of sex_code vector

As Functions: age_cat <- vectorize(function(x) { # x is input value
if (x < 18) { # condition
ret <- "< 18" # return label
} else if (x >= 18 & x < 24) {
ret <- "18 to 24"
} else if (x >= 24 & x < 45) {
ret <- "24 to 45"
} else if (x >= 45 & x < 60) {
ret <- "45 to 60"
} else if (x >= 60) {
ret <- "> 60"
} else {
ret <- "Unknown"
}
return(ret) })

df$age_cat <- age_cat(df$age) # apply function to age variable to create age_cat variable

proc means;

summarise(AllPages = sum(Pages),
AvgLength = mean(Pages),
AvgRating = mean(MyRating),
AvgReadTime = mean(read_time),
ShortRT = min(read_time),
LongRT = max(read_time),
TotalAuthors = n_distinct(Author))

proc contents;

library(Hmisc)

contents(dm)

7 types of dataframe compare results
1 - Ideal, Exact match
- For all variables and for all records, no difference in two datasets
2 - Inconsistent Sorting of Records
- Need to sort both dataframes by key variables, (Most important, First to correct)
3 - Misalignment of Key Variables using ID statement
- Need to correct variable name / type / length / label
4 - Record Count mismatch
- Need to view list of records in one dataframe but not in the other dataframe and update record selection condition / add / delete records as needed, test with subsets
5 - Variable mismatch
- Need to view the list of variables in one dataframe but not in the other dataframe and add / delete variables
6 - Data Values mismatch
- Need to view source dataframe with filter / sort to identify correct values if needed, update variable assignments, consider: case-sensitive, spaces, and rounding.
7 - Variable Attributes mismatch
- Need to update variable attributes (name / type / length / label) (Generally length and label mismatch may be acceptable, not not name or type)

# compare dataframes using diffdf

# diffdf( base=test_data , compare=test_data2 , keys = c("group1" , "group"))
# compare data frames by group1 and group 2 variables, useful to include key vars in differences
install.packages("diffdf")
library(diffdf)
LENGTH = 30 # assign variable value
suppressWarnings(RNGversion("3.5.0"))
set.seed(12334)
test_data <- tibble::tibble(
ID = 1:LENGTH,
GROUP1 = rep( c(1,2) , each = LENGTH/2),
GROUP2 = rep( c(1:(LENGTH/2)), 2 ),
INTEGER = rpois(LENGTH , 40),
BINARY = sample( c("M" , "F") , LENGTH , replace = T),
DATE = lubridate::ymd("2000-01-01") + rnorm(LENGTH, 0, 7000),
DATETIME = lubridate::ymd_hms("2000-01-01 00:00:00") + rnorm(LENGTH, 0, 200000000),
CONTINUOUS = rnorm(LENGTH , 30 , 12),
CATEGORICAL = factor(sample( c("A" , "B" , "C") , LENGTH , replace = T)), # factor variable type
LOGICAL = sample( c(TRUE , FALSE) , LENGTH , replace = T),
CHARACTER = stringi::stri_rand_strings(LENGTH, rpois(LENGTH , 13), pattern = "[ A-Za-z0-9]")
)
test_data # sample data frame
diffdf( test_data , test_data) # expects same variable names, type, label, order and number of rows
# first data frame is base and second data frame is compare
test_data2 <- test_data
test_data2 <- test_data2[,-6] # remove date variable
diffdf(test_data , test_data2 , suppress_warnings = T) # suppress warnings
test_data2 <- test_data
test_data2 <- test_data2[1:(nrow(test_data2) - 2),] # remove last two records
diffdf(test_data, test_data2 , suppress_warnings = T)
test_data2 <- test_data
test_data2[5, 2] <- 6 # assign group1 variable row 5 the value of 6
diffdf(test_data , test_data2 , suppress_warnings = T)
test_data2 <- test_data
test_data2[,2] <- as.character(test_data2[,2]) # change group1 from numeric to character variable type
diffdf(test_data , test_data2 , suppress_warnings = T)
test_data2 <- test_data
attr(test_data$ID , "label") <- "This is a interesting label"
attr(test_data2$ID , "label") <- "what do I type here?" # assign different label to ID
diffdf(test_data , test_data2 , suppress_warnings = T)
test_data2 <- test_data
levels(test_data2$CATEGORICAL) <- c(1,2,3) # assign different level values 1, 2 and 3 instead of A, B and C
diffdf(test_data , test_data2 , suppress_warnings = T)
test_data2 <- test_data
test_data2$INTEGER[c(5,2,15)] <- 99L # assign different value to variable
diffdf( test_data , test_data2 , keys = c("GROUP1" , "GROUP2") , suppress_warnings = T) # by group1 and group 2 variables, useful to include in differences
iris2 <- iris
for (i in 1:3) iris2[i,i] <- 99 # assign 99 values to iris2 data frame
diff <- diffdf( iris , iris2, suppress_warnings = TRUE) # save difference results to data frame
diffdf_issuerows( iris , diff) # display iris row issues
diffdf_issuerows( iris2 , diff) # display iris2 row issues
diffdf_issuerows( iris2 , diff , vars = "Sepal.Length")
diffdf_issuerows( iris2 , diff , vars = c("Sepal.Length" , "Sepal.Width"))
iris2 <- iris
for (i in 1:3) iris2[i,i] <- 99 # assign 99 values to iris2 data frame
diff <- diffdf( iris , iris2, suppress_warnings = TRUE)
diffdf_has_issues(diff) # true or false for data frame differences
if ( diffdf_has_issues(diff)){
#
}
dsin1 <- data.frame(x = 1.1e-06)
dsin2 <- data.frame(x = 1.1e-07) # assign different value
diffdf(dsin1 , dsin2 , suppress_warnings = T)
dsin1 <- data.frame(x = as.integer(c(1,2,3))) # integer variable
dsin2 <- data.frame(x = as.numeric(c(1,2,3))) # numeric variable
diffdf(dsin1 , dsin2 , suppress_warnings = T)
diffdf(dsin1 , dsin2 , suppress_warnings = T, strict_numeric = FALSE) # option to accept integer as numeric
dsin1 <- data.frame(x = as.character(c(1,2,3)), stringsAsFactors = FALSE)
dsin2 <- data.frame(x = as.factor(c(1,2,3))) # factors are numeric variable type
diffdf(dsin1 , dsin2 , suppress_warnings = T)
diffdf(dsin1 , dsin2 , suppress_warnings = T, strict_factor = FALSE) # option to accept character and factors