static void Main(string[] args)
{
var stopWatch = new Stopwatch();
stopWatch.Start();
Hashmap hashmap = new Hashmap();
string datafile = @"I:\Datascience\src\DataScience\Data\userItem.data";
string datafile2 = @"I:\Datascience\src\DataScience\Data\ml-latest-small\ratings.csv";
hashmap.Parsedata(datafile2, true);
// Console.WriteLine(Generalised(hashmap, 2, 1, 2, 31)); //1 = manhat 2 = eucli
//Console.WriteLine(Pearson(hashmap,1,2));
// var nn = NearestNeighbour(hashmap, 4, 0, 3);
// foreach (var neighbour in nn)
// {
// Console.WriteLine(neighbour.Key+":key Val:"+neighbour.Value);
// }
// Console.WriteLine(Predicted(nn, hashmap, 10));
var recs = TopNRec(hashmap, 8, 100, 0.35, 100, 5); //check if user rated more than 1 item
foreach (var i in recs)
{
Console.WriteLine(i.Key + ":key value: " + i.Value);
}
stopWatch.Stop();
Console.WriteLine("Elapsed in ms: " + stopWatch.ElapsedMilliseconds.ToString());
Console.ReadKey();
}
public static double Predicted(Dictionary <int, double> neighbours, Hashmap ratings, int item)
{
double c1 = 0;
double c2 = 0;
foreach (var kvpair in neighbours)
{
if (ratings.GetEntry(kvpair.Key).GetInnerDict().ContainsKey(item))
{
c1 += neighbours[kvpair.Key] * ratings.GetEntry(kvpair.Key).GetValueAtKey(item);
c2 += neighbours[kvpair.Key];
}
}
return(c1 / c2);
}
public static double Pearson(Hashmap ratings, int user_one, int user_two, Boolean nn)
{
double sx = 0, sy = 0, summ = 0, ssx = 0, ssy = 0;
var n = 0;
var user_one_ratings = ratings.GetEntry(user_one).GetInnerDict();
var user_two_ratings = ratings.GetEntry(user_two).GetInnerDict();
var similairRatingOne = new List <double>();
var similairRatingTwo = new List <double>();
foreach (var kvpair in user_one_ratings)
{
if (user_two_ratings.ContainsKey(kvpair.Key))
{
similairRatingOne.Add(user_one_ratings[kvpair.Key]);
similairRatingTwo.Add(user_two_ratings[kvpair.Key]);
}
}
if (similairRatingOne.Count == similairRatingTwo.Count)
{
n = similairRatingOne.Count;
}
if (n == 0 && nn == false)
{
throw new Exception("No items for users");
}
for (var i = 0; i < n; i++)
{
summ += similairRatingOne[i] * similairRatingTwo[i];
sx += similairRatingOne[i];
sy += similairRatingTwo[i];
ssx += Pow(similairRatingOne[i], 2);
ssy += Pow(similairRatingTwo[i], 2);
}
var c1 = summ - sx * sy / n;
var c2 = Sqrt(ssx - Pow(sx, 2) / n);
var c3 = Sqrt(ssy - Pow(sy, 2) / n);
return(c1 / (c2 * c3));
}
public static Hashmap ParseLinq(string datapath)
{
Hashmap generatedDict = new Hashmap();
// Step 1: Data
var lines = File.ReadLines(datapath);
// Step 2: Query
var linqDic = lines
.Select(line => line.Split(','))
.Select(entry => new { UserId = Int16.Parse(entry[0]), Article = Int16.Parse(entry[1]), Rating = Decimal.Parse(entry[2]) })
;
// Step 3: Execution
foreach (string line in lines)
{
}
return(generatedDict);
}
public static Dictionary <int, double> NearestNeighbour(Hashmap ratings, int target, double treshhold, int K)
{
var inner = ratings.GetInner();
var nearest = new Dictionary <int, double>();
foreach (var kvpair in inner)
{
if (kvpair.Key != target)
{
var pCo = Pearson(ratings, kvpair.Key, target, true);
if (pCo >= treshhold)
{
nearest.Add(kvpair.Key, pCo);
}
}
}
var sortedNeighbours = nearest.OrderByDescending(x => x.Value).Take(K).ToDictionary(pair => pair.Key, pair => pair.Value);
return(sortedNeighbours);
}
public static double Generalised(Hashmap ratings, int r, int user_one, int user_two, int item)
{
var user_one_ratings = ratings.GetEntry(user_one).GetInnerDict();
var user_two_ratings = ratings.GetEntry(user_two).GetInnerDict();
double sums = 0;
if (user_two_ratings.ContainsKey(item) && user_one_ratings.ContainsKey(item))
{
sums += Pow(Abs(user_one_ratings[item] - user_two_ratings[item]), r);
}
else
{
throw new Exception("No item for users");
}
var answer = Pow(sums, 1 / r);
return(Similarity(answer));
}
public static Dictionary <int, double> TopNRec(Hashmap ratings, int user, int N, double treshhold, int K, int same)
{
var userMovieId = new List <int>();
var nnMovieId = new List <int>();
var nn = NearestNeighbour(ratings, user, treshhold, K);
var rec = new Dictionary <int, double>();
foreach (var kvpair in ratings.GetInner())
{
if (kvpair.Key == user)
{
foreach (var innerpair in kvpair.Value.GetInnerDict())
{
userMovieId.Add(innerpair.Key);
}
}
}
foreach (var kvpair in ratings.GetInner())
{
foreach (var innerpair in kvpair.Value.GetInnerDict())
{
if (!userMovieId.Contains(innerpair.Key))
{
nnMovieId.Add(innerpair.Key);
}
}
}
var moviesIds = nnMovieId.GroupBy(x => x).Where(group => group.Count() > same - 1).Select(group => group.Key);
foreach (var movie in moviesIds)
{
rec.Add(movie, Predicted(nn, ratings, movie));
}
var topN = rec.OrderByDescending(x => x.Value).Take(N).ToDictionary(pair => pair.Key, pair => pair.Value);
return(topN);
}
public static double Cosine(Hashmap ratings, int user_one, int user_two)
{
double summ = 0, ssx = 0, ssy = 0;
var n = 0;
var user_one_ratings = ratings.GetEntry(user_one).GetInnerDict();
var user_two_ratings = ratings.GetEntry(user_two).GetInnerDict();
var similairRatingOne = new List <double>();
var similairRatingTwo = new List <double>();
foreach (var kvpair in user_one_ratings)
{
if (user_two_ratings.ContainsKey(kvpair.Key) && user_one_ratings.ContainsKey(kvpair.Key))
{
similairRatingOne.Add(user_one_ratings[kvpair.Key]);
similairRatingTwo.Add(user_two_ratings[kvpair.Key]);
}
else
{
throw new Exception("No sim");
}
}
if (similairRatingOne.Count == similairRatingTwo.Count)
{
n = similairRatingOne.Count;
}
for (var i = 0; i < n; i++)
{
summ += similairRatingOne[i] * similairRatingTwo[i];
ssx += Pow(similairRatingOne[i], 2);
ssy += Pow(similairRatingTwo[i], 2);
}
return(summ / (Sqrt(ssx) * Sqrt(ssy)));
}
