private List<DataEntry> getMatchesFromYear(String country, String division, int year) {
List<DataEntry> retval = new LinkedList<DataEntry>();
List<DataEntry> tmp = Model.dataset.get(this.getDataKey(country, division));
for (DataEntry tse : tmp) {
if (year == Integer.valueOf((String) tse.getExtra("season_start_year")))
retval.add(tse.clone());
}
return retval;
}
private List<DataEntry> getTrainingSet(String country, String division) {
country = country.toLowerCase();
List<DataEntry> retval = new LinkedList<DataEntry>();
List<DataEntry> tmp = dataset.get(this.getDataKey(country, division));
for (DataEntry tse : tmp) {
retval.add(tse.clone());
}
return retval;
}
private List<DataEntry> getTrainingSet(String country, String division, int maxyear) {
List<DataEntry> retval = new LinkedList<DataEntry>();
List<DataEntry> clone = dataset.get(this.getDataKey(country, division));
for (DataEntry tse : clone) {
if (maxyear >= Integer.valueOf((String) tse.getExtra("season_start_year")))
retval.add(tse.clone());
}
return retval;
}
private void printDatasetToWekaFormat(Map<String, LinkedList<DataEntry>> data, boolean recurse)
throws IOException {
if (recurse) {
for (String country : data.keySet()) {
Map<String, LinkedList<DataEntry>> ndata = new TreeMap<String, LinkedList<DataEntry>>();
ndata.put(country, data.get(country));
printDatasetToWekaFormat(ndata, false);
}
} else {
String key = "";
for (String s : data.keySet()) key = s;
String[] split = key.split("_");
List<DataEntry> trainingset = this.getTrainingSet(split[0], split[1], 2012);
List<DataEntry> matchset = this.getMatchesFromYear(split[0], split[1], 2012);
trainingset.addAll(matchset); // Add the match set
Writer writer =
new BufferedWriter(
new OutputStreamWriter(
new FileOutputStream(Setup.outputpath + key + ".arff"), "utf-8"));
writer.write("@relation " + key + "\r\n");
writer.write("@attribute 'rating home' real\r\n");
writer.write("@attribute 'rating away' real\r\n");
writer.write("@attribute 'current run hjemme' real\r\n");
writer.write("@attribute 'current run ude' real\r\n");
writer.write("@attribute 'run hjemme' real\r\n");
writer.write("@attribute 'run ude' real\r\n");
writer.write("@attribute 'class' {Hjemme, Uafgjort, Ude}\r\n");
writer.write("@data\r\n");
// And loop
for (DataEntry tse : trainingset) {
writer.write(
tse.getX()[0]
+ ","
+ tse.getX()[1]
+ ","
+ tse.getX()[2]
+ ","
+ tse.getX()[3]
+ ","
+ tse.getX()[4]
+ ","
+ tse.getX()[5]
+ ","
+ tse.getY()
+ "\r\n");
}
writer.close();
}
}
public synchronized void load() throws SQLException {
// And calculate
Connection connection = getDb();
Statement stmt = connection.createStatement();
ResultSet scoreSet = stmt.executeQuery("SELECT hold,division,land from klubber");
while (scoreSet.next()) {
// Fetch info
String land = scoreSet.getString("land");
String hold = scoreSet.getString("hold");
int division = scoreSet.getInt("division");
double rating = Setup.pointStart / (Math.pow(Setup.divisionsforskel, division));
Team team = new Team(hold, land, division, rating, Setup.matchesToStore);
teams.addTeam(team);
}
stmt.close();
// Load the matches
stmt = connection.createStatement();
String matchesSQL = "SELECT * FROM kampe order by dato asc";
ResultSet r = stmt.executeQuery(matchesSQL);
while (r.next()) {
try {
// Fetch info
String country = r.getString("land");
Team away = teams.get(country, r.getString("udehold"));
int division = r.getInt("division");
int goalsAway = r.getInt("udemaal");
int goalsHome = r.getInt("hjemmemaal");
String seasonStartYear = r.getString("season_start_year");
Team home = teams.get(country, r.getString("hjemmehold"));
int round = r.getInt("runde");
Kamp match = new Kamp(home, away, goalsHome, goalsAway, round, seasonStartYear);
// Find result
Udfald y = null;
if (goalsHome > goalsAway) y = Udfald.HJEMME;
else if (goalsHome == goalsAway) y = Udfald.UAFGJORT;
else y = Udfald.UDE;
double[] features = getKamp(home, away, country);
DataEntry matchInfo = new DataEntry(features, y);
matchInfo.addExtra("season_start_year", seasonStartYear);
String key = getDataKey(country, division + "");
synchronized (this) {
LinkedList<DataEntry> matchData = dataset.get(key);
if (matchData == null) {
matchData = new LinkedList<DataEntry>();
dataset.put(key, matchData);
}
// If enough matches and so on, update
if (home.totalplayed >= Setup.TRAINING_ROUNDS
&& away.totalplayed >= Setup.TRAINING_ROUNDS // Total played at least
&& round >= Setup.antalRunder // Total played in start of each season
) {
matchData.add(matchInfo);
}
}
// And update a lot of information
// Skal altid ske. Til træning af klubbernes point
teams.opdaterPoint(home, away, goalsHome, goalsAway, division);
// The rest
home.addLatestHomeMatch(match);
home.addLatestMatch(match);
away.addLatestAwayMatch(match);
away.addLatestMatch(match);
home.addInternal(match);
away.addInternal(match);
home.totalplayed++;
away.totalplayed++;
} catch (Exception e) {
Log.e(e);
}
}
r.close();
stmt.close();
// Filter out noise if needed
if (Setup.FILTER_DATA) {
for (String key : dataset.keySet()) {
LinkedList<DataEntry> matches = dataset.get(key);
Setup.FILTER_HANDLER.applyFilters(matches);
}
}
// Prepare normalizer
for (String key : dataset.keySet()) {
Normalizer normalizer = new Normalizer(dataset.get(key), Normalizer.Method.NORMALIZATION);
normalizers.put(key, normalizer);
if (Setup.NORMALIZE_DATA) normalizer.apply(dataset.get(key));
}
}