public void loadData(Review[] reviews) {
for (Review aReview : reviews) {
UserVector uv = users.get(aReview.user_id);
if (uv == null) {
uv = new UserVector();
}
uv.addReview(aReview.business_id, aReview.rating);
users.put(aReview.user_id, uv);
}
}
public double[] getAvgRatings(Cluster c) {
double[] ratings = new double[dimensions];
double[] counter = new double[dimensions];
for (UserVector user : c.getUsers().values()) {
HashMap<Integer, Double> vect = user.getVector();
for (Entry<Integer, Double> e : vect.entrySet()) {
ratings[e.getKey()] += e.getValue();
++counter[e.getKey()];
}
}
// Average Total Ratings
for (int idx = 0; idx < dimensions; ++idx) {
if (counter[idx] != 0) {
ratings[idx] /= counter[idx];
}
}
return ratings;
}
public void performKmeans(int K) {
for (int i = 0; i < K; i++) {
Cluster aCluster = new Cluster(dimensions);
clusters.add(aCluster);
}
boolean done = false;
int numberOfSteps = 0;
while (!done && (numberOfSteps < 20)) {
System.out.println("Step #" + (++numberOfSteps));
for (Cluster c : clusters) {
c.clearUsers();
}
done = true;
for (int idx = 0; idx < users.size(); ++idx) {
UserVector user = users.get(idx);
Cluster closestCluster = null;
double minDistance = 1000000.0;
for (Cluster myCluster : clusters) {
double[] myClusterCenter = myCluster.getCenter();
double[] userVect = user.toArray(dimensions);
double dist = getDistance(userVect, myClusterCenter);
if (dist < minDistance) {
minDistance = dist;
closestCluster = myCluster;
}
}
closestCluster.addUser(new Integer(idx), user);
}
for (Cluster c : clusters) {
done = done & c.calculateCenter();
}
}
}