main_marking2/danube_test6.scala
author Christian Urban <christian.urban@kcl.ac.uk>
Sun, 01 May 2022 12:41:56 +0100
changeset 425 957808dcb367
parent 424 daf561a83ba6
permissions -rw-r--r--
updated
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
389
4113d4d8cf62 updated marking 2
Christian Urban <christian.urban@kcl.ac.uk>
parents:
diff changeset
     1
424
Christian Urban <christian.urban@kcl.ac.uk>
parents: 389
diff changeset
     2
import M2._
389
4113d4d8cf62 updated marking 2
Christian Urban <christian.urban@kcl.ac.uk>
parents:
diff changeset
     3
4113d4d8cf62 updated marking 2
Christian Urban <christian.urban@kcl.ac.uk>
parents:
diff changeset
     4
// first test 
4113d4d8cf62 updated marking 2
Christian Urban <christian.urban@kcl.ac.uk>
parents:
diff changeset
     5
4113d4d8cf62 updated marking 2
Christian Urban <christian.urban@kcl.ac.uk>
parents:
diff changeset
     6
4113d4d8cf62 updated marking 2
Christian Urban <christian.urban@kcl.ac.uk>
parents:
diff changeset
     7
def urban_groupById(ratings: List[(String, String)]) = 
4113d4d8cf62 updated marking 2
Christian Urban <christian.urban@kcl.ac.uk>
parents:
diff changeset
     8
  ratings.groupBy(_._1).view.mapValues(_.map(_._2)).toMap 
4113d4d8cf62 updated marking 2
Christian Urban <christian.urban@kcl.ac.uk>
parents:
diff changeset
     9
4113d4d8cf62 updated marking 2
Christian Urban <christian.urban@kcl.ac.uk>
parents:
diff changeset
    10
def urban_get_csv_file(name: String) : List[String] = {
4113d4d8cf62 updated marking 2
Christian Urban <christian.urban@kcl.ac.uk>
parents:
diff changeset
    11
  import io.Source
4113d4d8cf62 updated marking 2
Christian Urban <christian.urban@kcl.ac.uk>
parents:
diff changeset
    12
  import scala.util._
4113d4d8cf62 updated marking 2
Christian Urban <christian.urban@kcl.ac.uk>
parents:
diff changeset
    13
  val csv = Source.fromFile(name)
4113d4d8cf62 updated marking 2
Christian Urban <christian.urban@kcl.ac.uk>
parents:
diff changeset
    14
  csv.mkString.split("\n").toList.drop(1)
4113d4d8cf62 updated marking 2
Christian Urban <christian.urban@kcl.ac.uk>
parents:
diff changeset
    15
}
4113d4d8cf62 updated marking 2
Christian Urban <christian.urban@kcl.ac.uk>
parents:
diff changeset
    16
4113d4d8cf62 updated marking 2
Christian Urban <christian.urban@kcl.ac.uk>
parents:
diff changeset
    17
def urban_process_ratings(lines: List[String]) : List[(String, String)] = {
4113d4d8cf62 updated marking 2
Christian Urban <christian.urban@kcl.ac.uk>
parents:
diff changeset
    18
  for (cols <- lines.map(_.split(",").toList); 
4113d4d8cf62 updated marking 2
Christian Urban <christian.urban@kcl.ac.uk>
parents:
diff changeset
    19
       if (cols(2).toFloat >= 4)) yield (cols(0), cols(1))  
4113d4d8cf62 updated marking 2
Christian Urban <christian.urban@kcl.ac.uk>
parents:
diff changeset
    20
}
4113d4d8cf62 updated marking 2
Christian Urban <christian.urban@kcl.ac.uk>
parents:
diff changeset
    21
4113d4d8cf62 updated marking 2
Christian Urban <christian.urban@kcl.ac.uk>
parents:
diff changeset
    22
def urban_process_movies(lines: List[String]) : List[(String, String)] = {
4113d4d8cf62 updated marking 2
Christian Urban <christian.urban@kcl.ac.uk>
parents:
diff changeset
    23
  for (cols <- lines.map(_.split(",").toList)) yield (cols(0), cols(1))  
4113d4d8cf62 updated marking 2
Christian Urban <christian.urban@kcl.ac.uk>
parents:
diff changeset
    24
}
4113d4d8cf62 updated marking 2
Christian Urban <christian.urban@kcl.ac.uk>
parents:
diff changeset
    25
4113d4d8cf62 updated marking 2
Christian Urban <christian.urban@kcl.ac.uk>
parents:
diff changeset
    26
4113d4d8cf62 updated marking 2
Christian Urban <christian.urban@kcl.ac.uk>
parents:
diff changeset
    27
val urban_good_ratings = process_ratings(urban_get_csv_file("ratings.csv"))
4113d4d8cf62 updated marking 2
Christian Urban <christian.urban@kcl.ac.uk>
parents:
diff changeset
    28
val urban_movie_names = process_movies(urban_get_csv_file("movies.csv")).toMap
4113d4d8cf62 updated marking 2
Christian Urban <christian.urban@kcl.ac.uk>
parents:
diff changeset
    29
4113d4d8cf62 updated marking 2
Christian Urban <christian.urban@kcl.ac.uk>
parents:
diff changeset
    30
val urban_ratings_map = urban_groupById(urban_good_ratings)
4113d4d8cf62 updated marking 2
Christian Urban <christian.urban@kcl.ac.uk>
parents:
diff changeset
    31
4113d4d8cf62 updated marking 2
Christian Urban <christian.urban@kcl.ac.uk>
parents:
diff changeset
    32
assert((for (n <- List("1", "2", "3", "4", "5")) yield {
4113d4d8cf62 updated marking 2
Christian Urban <christian.urban@kcl.ac.uk>
parents:
diff changeset
    33
  recommendations(urban_ratings_map, urban_movie_names, n).length
4113d4d8cf62 updated marking 2
Christian Urban <christian.urban@kcl.ac.uk>
parents:
diff changeset
    34
}) == List(2, 2, 2, 0, 2))
4113d4d8cf62 updated marking 2
Christian Urban <christian.urban@kcl.ac.uk>
parents:
diff changeset
    35