Last active
June 2, 2018 14:29
-
-
Save jlln/51c5bb1a1d8ec2f6956d to your computer and use it in GitHub Desktop.
One-hot encoder for use with Spark DataFrames.
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
import scala.collection.JavaConverters._ | |
import org.apache.spark.sql.types.{StructType,StructField,StringType} | |
import org.apache.spark.sql.Row | |
def identityMatrix(n:Int):Array[Array[String]]=Array.tabulate(n,n)((x,y) => if(x==y) "1" else "0") | |
def encodeStringOneHot(table:org.apache.spark.sql.DataFrame,column:String) = { | |
//Accepts the dataframe and the target column name. Returns a new dataframe in which the target column has been replaced with a one-hot/dummy encoding. | |
table.registerTempTable("temp") | |
val categories_table = sqlContext.sql(s"SELECT DISTINCT $column FROM temp") | |
val n_categories:Int = categories_table.count().toInt | |
val categories = categories_table.collectAsList().asScala.map(_.get(0).toString).toList | |
val matrix = categories.zip(identityMatrix(n_categories)).map{ | |
case (c,r) => c+:r | |
} | |
val matrix_rdd = sc.makeRDD(matrix) | |
val schema = StructType(StructField(column,StringType,true) +: categories.map(c=> StructField(c,StringType,true))) | |
val row_rdd:RDD[Row] = matrix_rdd.map(r => Row.fromSeq(r)) | |
val table_oh = sqlContext.createDataFrame(row_rdd,schema) | |
var joined = table.join(table_oh,List(column),"left_outer").drop(column) | |
categories.map{ | |
c=>{ | |
joined = joined.withColumnRenamed(c,column+"_"+c) | |
} | |
} | |
joined | |
} |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment