4 object CW6b { |
4 object CW6b { |
5 |
5 |
6 import io.Source |
6 import io.Source |
7 import scala.util._ |
7 import scala.util._ |
8 |
8 |
9 def get_csv_page(url: String) : List[String] = |
|
10 Source.fromURL(url)("ISO-8859-1").getLines.toList |
|
11 |
|
12 def get_csv_file(file: String) : List[String] = |
|
13 Source.fromFile(file)("ISO-8859-1").getLines.toList |
|
14 |
|
15 |
|
16 val url_alcohol = |
9 val url_alcohol = |
17 "https://raw.githubusercontent.com/fivethirtyeight/data/master/alcohol-consumption/drinks.csv" |
10 "https://raw.githubusercontent.com/fivethirtyeight/data/master/alcohol-consumption/drinks.csv" |
18 |
11 |
19 val file_population = |
12 val file_population = |
20 "population.csv" |
13 "population.csv" |
21 |
14 |
22 get_csv_page(url_alcohol) |
|
23 get_csv_file(file_population) |
|
24 |
15 |
25 get_csv_page(url_alcohol).size |
16 //(1) Complete the get_csv_page function below. It takes the URL-string |
26 get_csv_file(file_population).size |
17 // as argument and generates a list of strings corresponing to each |
|
18 // line in the csv list. The URL url_alcohol is one possible argument. |
27 |
19 |
28 val alcs = get_csv_page(url_alcohol) |
20 //def get_csv_page(url: String) : List[String] = ... |
29 val pops = get_csv_file(file_population) |
|
30 |
21 |
31 |
22 |
32 def process_alcs(lines: List[String]) : List[(String, Double)] = |
23 // Complete the get_csv_file function below. It takes a file-string |
33 for (l <- lines) yield { |
24 // as argument and reads the content of the given file. Like above |
34 val entries = l.split(",").toList |
25 // it should generates a list of strings corresponing to each |
35 (entries(0), entries(4).toDouble) |
26 // line in the csv-list. The filename file_population is one possible |
36 } |
27 // argument. |
37 |
28 |
38 def process_pops(lines: List[String]) : Map[String, Long] = |
29 //def get_csv_file(file: String) : List[String] = ... |
39 (for (l <- lines) yield { |
|
40 val entries = l.split(",").toList |
|
41 (entries(0), entries(1).toLong) |
|
42 }).toMap |
|
43 |
30 |
44 |
31 |
45 def sorted_country_consumption() : List[(String, Long)] = { |
|
46 val alcs2 = process_alcs(alcs.drop(1)) |
|
47 val pops2 = process_pops(pops.drop(1)) |
|
48 val cons_list = |
|
49 for ((cname, cons) <- alcs2; |
|
50 if pops2.isDefinedAt(cname)) yield (cname, (cons * pops2(cname)).toLong) |
|
51 cons_list.sortBy(_._2).reverse |
|
52 } |
|
53 |
32 |
54 sorted_country_consumption().take(10) |
33 //(2) Complete the functions that process the csv-lists. For |
55 sorted_country_consumption().size |
34 // process_alcs extract the country name (as String) and the |
|
35 // pure alcohol consumption (as Double). For process_pops |
|
36 // generate a Map of Strings (country names) to Long numbers |
|
37 // (population size). |
56 |
38 |
57 def percentage(n: Int) : (Long, Long, Double) = { |
39 //def process_alcs(lines: List[String]) : List[(String, Double)] = ... |
58 val cons_list = sorted_country_consumption() |
40 |
59 val sum_n = cons_list.take(n).map(_._2).sum |
41 //def process_pops(lines: List[String]) : Map[String, Long] = ... |
60 val sum_all = cons_list.map(_._2).sum |
|
61 val perc = (sum_n.toDouble / sum_all.toDouble) * 100.0 |
|
62 (sum_all, sum_n, perc) |
|
63 } |
|
64 |
42 |
65 |
43 |
66 percentage(10) |
44 |
67 percentage(164) |
45 //(3) Calculate for each country the overall alcohol_consumption using |
|
46 // the data from the alcohol list and the population sizes list. You |
|
47 // should only include countries on the alcohol list that are also |
|
48 // (with the exact name) on the population sizes list. Note that |
|
49 // the spelling of some names in the alcohol list differs from the |
|
50 // population sizes list. Sort the resulting list according to the |
|
51 // country with the highest alcohol consumption to the country |
|
52 // with the lowest alcohol consumption. |
|
53 |
|
54 //def sorted_country_consumption() : List[(String, Long)] = ... |
|
55 |
|
56 |
|
57 // Calculate the world consumption of pure alcohol of all countries, which |
|
58 // should be the first element in the tuple below. The second element is |
|
59 // the overall consumption of the first n countries in the sorted list |
|
60 // from above; and finally the double should be the percentage of the |
|
61 // first n countries of the the world consumption of alcohol. |
|
62 |
|
63 //def percentage(n: Int) : (Long, Long, Double) = ... |
68 |
64 |
69 } |
65 } |