R Language Spark API (SparkR) Create RDDs (Resilient Distributed Datasets)


From dataframe:

mtrdd <- createDataFrame(sqlContext, mtcars)

From csv:

For csv's, you need to add the csv package to the environment before initiating the Spark context:

Sys.setenv('SPARKR_SUBMIT_ARGS'='"--packages" "com.databricks:spark-csv_2.10:1.4.0" "sparkr-shell"') # context for csv import read csv -> 
sc <- sparkR.init()
sqlContext <- sparkRSQL.init(sc)

Then, you can load the csv either by infering the data schema of the data in the columns:

train <- read.df(sqlContext, "/train.csv", header= "true", source = "com.databricks.spark.csv", inferSchema = "true")

Or by specifying the data schema beforehand:

 customSchema <- structType(
    structField("margin", "integer"),
    structField("gross", "integer"),
    structField("name", "string"))

 train <- read.df(sqlContext, "/train.csv", header= "true", source = "com.databricks.spark.csv", schema = customSchema)