|
1 // Core Part about Movie Recommendations |
|
2 // at Danube.co.uk |
|
3 //=========================================== |
|
4 |
|
5 import io.Source |
|
6 import scala.util._ |
|
7 |
|
8 object CW7b { // for purposes of generating a jar |
|
9 |
|
10 // (1) Implement the function get_csv_url which takes an url-string |
|
11 // as argument and requests the corresponding file. The two urls |
|
12 // of interest are ratings_url and movies_url, which correspond |
|
13 // to CSV-files. |
|
14 // The function should ReTurn the CSV file appropriately broken |
|
15 // up into lines, and the first line should be dropped (that is without |
|
16 // the header of the CSV file). The result is a list of strings (lines |
|
17 // in the file). |
|
18 |
|
19 def get_csv_url(url: String) : List[String] = { |
|
20 val csv = Source.fromURL(url)("ISO-8859-1") |
|
21 csv.mkString.split("\n").toList.drop(1) |
|
22 } |
|
23 |
|
24 val ratings_url = """https://nms.kcl.ac.uk/christian.urban/ratings.csv""" |
|
25 val movies_url = """https://nms.kcl.ac.uk/christian.urban/movies.csv""" |
|
26 |
|
27 // test cases |
|
28 |
|
29 //val ratings = get_csv_url(ratings_url) |
|
30 //val movies = get_csv_url(movies_url) |
|
31 |
|
32 //ratings.length // 87313 |
|
33 //movies.length // 9742 |
|
34 |
|
35 // (2) Implement two functions that process the CSV files. The ratings |
|
36 // function filters out all ratings below 4 and ReTurns a list of |
|
37 // (userID, movieID) pairs. The movies function just ReTurns a list |
|
38 // of (movieId, title) pairs. |
|
39 |
|
40 |
|
41 def process_ratings(lines: List[String]) : List[(String, String)] = { |
|
42 for (cols <- lines.map(_.split(",").toList); |
|
43 if (cols(2).toFloat >= 4)) yield (cols(0), cols(1)) |
|
44 } |
|
45 |
|
46 def process_movies(lines: List[String]) : List[(String, String)] = { |
|
47 for (cols <- lines.map(_.split(",").toList)) yield (cols(0), cols(1)) |
|
48 } |
|
49 |
|
50 // test cases |
|
51 |
|
52 //val good_ratings = process_ratings(ratings) |
|
53 //val movie_names = process_movies(movies) |
|
54 |
|
55 //good_ratings.length //48580 |
|
56 //movie_names.length // 9742 |
|
57 |
|
58 //============================================== |
|
59 // Do not change anything below, unless you want |
|
60 // to submit the file for the advanced part 3! |
|
61 //============================================== |
|
62 |
|
63 |
|
64 // (3) Implement a grouping function that calulates a map |
|
65 // containing the userIds and all the corresponding recommendations |
|
66 // (list of movieIds). This should be implemented in a tail |
|
67 // recursive fashion, using a map m as accumulator. This map |
|
68 // is set to Map() at the beginning of the claculation. |
|
69 |
|
70 def groupById(ratings: List[(String, String)], |
|
71 m: Map[String, List[String]]) : Map[String, List[String]] = ratings match { |
|
72 case Nil => m |
|
73 case (id, mov) :: rest => { |
|
74 val old_ratings = m.getOrElse (id, Nil) |
|
75 val new_ratings = m + (id -> (mov :: old_ratings)) |
|
76 groupById(rest, new_ratings) |
|
77 } |
|
78 } |
|
79 |
|
80 // |
|
81 //val ls = List(("1", "a"), ("2", "a"), ("1", "c"), ("2", "a"), ("1", "c")) |
|
82 // |
|
83 //val m = groupById(ls, Map()) |
|
84 // |
|
85 //m.getOrElse("1", Nil).count(_ == "c") // => 2 |
|
86 //m.getOrElse("1", Nil).count(_ == "a") // => 1 |
|
87 |
|
88 // test cases |
|
89 //val ratings_map = groupById(good_ratings, Map()) |
|
90 //groupById(good_ratings, Map()).get("214") |
|
91 //groupById(good_ratings, Map()).toList.minBy(_._2.length) |
|
92 //val movies_map = movie_names.toMap |
|
93 |
|
94 //ratings_map.get("414").get.map(movies_map.get(_)) // most prolific recommender with 1227 positive ratings |
|
95 //ratings_map.get("474").get.map(movies_map.get(_)) // second-most prolific recommender with 787 positive ratings |
|
96 //ratings_map.get("214").get.map(movies_map.get(_)) // least prolific recommender with only 1 positive rating |
|
97 |
|
98 |
|
99 |
|
100 //(4) Implement a function that takes a ratings map and a movie_name as argument. |
|
101 // The function calculates all suggestions containing |
|
102 // the movie mov in its recommendations. It ReTurns a list of all these |
|
103 // recommendations (each of them is a list and needs to have mov deleted, |
|
104 // otherwise it might happen we recommend the same movie). |
|
105 |
|
106 def favourites(m: Map[String, List[String]], mov: String) : List[List[String]] = |
|
107 (for (id <- m.keys.toList; |
|
108 if m(id).contains(mov)) yield m(id).filter(_ != mov)) |
|
109 |
|
110 |
|
111 |
|
112 // test cases |
|
113 // movie ID "912" -> Casablanca (1942) |
|
114 // "858" -> Godfather |
|
115 // "260" -> Star Wars: Episode IV - A New Hope (1977) |
|
116 |
|
117 //favourites(ratings_map, "912").length // => 80 |
|
118 |
|
119 // That means there are 80 users that recommend the movie with ID 912. |
|
120 // Of these 80 users, 55 gave a good rating to movie 858 and |
|
121 // 52 a good rating to movies 260, 318, 593. |
|
122 |
|
123 |
|
124 // (5) Implement a suggestions function which takes a rating |
|
125 // map and a movie_name as arguments. It calculates all the recommended |
|
126 // movies sorted according to the most frequently suggested movie(s) first. |
|
127 |
|
128 // needed in Scala 2.13. |
|
129 |
|
130 def mapValues[S, T, R](m: Map[S, T], f: T => R) = |
|
131 m.map { case (x, y) => (x, f(y)) } |
|
132 |
|
133 def suggestions(recs: Map[String, List[String]], |
|
134 mov_name: String) : List[String] = { |
|
135 val favs = favourites(recs, mov_name).flatten |
|
136 val favs_counted = mapValues(favs.groupBy(identity), (v:List[String]) => v.size).toList |
|
137 val favs_sorted = favs_counted.sortBy(_._2).reverse |
|
138 favs_sorted.map(_._1) |
|
139 } |
|
140 |
|
141 // check |
|
142 // groupMap is equivalent to groupBy(key).mapValues(_.map(f)) |
|
143 |
|
144 // test cases |
|
145 |
|
146 //suggestions(ratings_map, "912") |
|
147 //suggestions(ratings_map, "912").length |
|
148 // => 4110 suggestions with List(858, 260, 318, 593, ...) |
|
149 // being the most frequently suggested movies |
|
150 |
|
151 // (6) Implement recommendations functions which generates at most |
|
152 // *two* of the most frequently suggested movies. It Returns the |
|
153 // actual movie names, not the movieIDs. |
|
154 |
|
155 def recommendations(recs: Map[String, List[String]], |
|
156 movs: Map[String, String], |
|
157 mov_name: String) : List[String] = |
|
158 suggestions(recs, mov_name).take(2).map(movs.get(_).get) |
|
159 |
|
160 |
|
161 // testcases |
|
162 |
|
163 // recommendations(ratings_map, movies_map, "912") |
|
164 // => List(Godfather, Star Wars: Episode IV - A NewHope (1977)) |
|
165 |
|
166 //recommendations(ratings_map, movies_map, "260") |
|
167 // => List(Star Wars: Episode V - The Empire Strikes Back (1980), |
|
168 // Star Wars: Episode VI - Return of the Jedi (1983)) |
|
169 |
|
170 // recommendations(ratings_map, movies_map, "2") |
|
171 // => List(Lion King, Jurassic Park (1993)) |
|
172 |
|
173 // recommendations(ratings_map, movies_map, "0") |
|
174 // => Nil |
|
175 |
|
176 // recommendations(ratings_map, movies_map, "1") |
|
177 // => List(Shawshank Redemption, Forrest Gump (1994)) |
|
178 |
|
179 // recommendations(ratings_map, movies_map, "4") |
|
180 // => Nil (there are three ratings fro this movie in ratings.csv but they are not positive) |
|
181 |
|
182 |
|
183 // If you want to calculate the recomendations for all movies. |
|
184 // Will take a few seconds calculation time. |
|
185 |
|
186 //val all = for (name <- movie_names.map(_._1)) yield { |
|
187 // recommendations(ratings_map, movies_map, name) |
|
188 //} |
|
189 |
|
190 // helper functions |
|
191 //List().take(2) |
|
192 //List(1).take(2) |
|
193 //List(1,2).take(2) |
|
194 //List(1,2,3).take(2) |
|
195 |
|
196 } |