public static void svmTrain(ArrayList<ArrayList<Double>> train) {
svm_problem prob = new svm_problem();
int dataCount = train.size();
prob.y = new double[dataCount];
prob.l = dataCount;
prob.x = new svm_node[dataCount][];
for (int i = 0; i < dataCount; i++) {
ArrayList<Double> features = train.get(i);
prob.x[i] = new svm_node[features.size() - 1];
for (int j = 1; j < features.size(); j++) {
svm_node node = new svm_node();
node.index = j - 1;
node.value = features.get(j);
prob.x[i][j - 1] = node;
}
prob.y[i] = features.get(0);
}
svm_parameter param = new svm_parameter();
param.probability = 1;
param.gamma = 0.5;
param.nu = 0.5;
param.C = 1;
param.svm_type = svm_parameter.NU_SVC;
param.kernel_type = svm_parameter.LINEAR;
param.cache_size = 20000;
param.eps = 0.001;
_model = svm.svm_train(prob, param);
}
public void addPrev(ArrayList<MultiPoint> data, svm_node[] buf, int i) {
int L = data.get(0).values.length;
for (int j = 0; j < prev; j++) {
for (int l = 0; l < L; l++) {
svm_node node = new svm_node();
node.index = colCnt + j;
node.value = data.get(i - j - 1).values[l];
buf[colCnt + j * L + l] = node;
}
}
// svm_node node = new svm_node();
// node.index = colCnt+prev;
// node.value = 1.0;
// buf[colCnt+prev*L]=node;
colCnt += prev * L;
}
public static LabelAndProbability evaluate(ArrayList<Double> features) {
svm_node[] nodes = new svm_node[features.size() - 1];
for (int i = 1; i < features.size(); i++) {
svm_node node = new svm_node();
node.index = i;
node.value = features.get(i);
nodes[i - 1] = node;
}
int[] labels = new int[Globals._numExercises];
int n = Globals._numExercises;
Log.d("Number of Exercises", "" + n);
double[] values = new double[(n * (n - 1) / 2)];
svm.svm_get_labels(_model, labels);
for (int i = 0; i < n; i++) {
Log.d("labels", "" + labels[i]);
}
int Label = (int) svm.svm_predict_values(_model, nodes, values);
double accuracy = 0.0;
int votes = 0;
int p = 0;
for (int i = 0; i < n; i++)
for (int j = i + 1; j < n; j++) {
if (Label == labels[i]) {
accuracy += values[p];
if (values[p] > 0) votes++;
} else if (Label == labels[j]) {
accuracy -= values[p];
if (values[p] < 0) votes++;
}
p++;
}
// int label = 0;
// for (int i = 0; i < probabilities.length; i++) {
// Log.d("Labels and Probabilites", labels[i] + " : " + probabilities[i]);
// if (max < probabilities[i]) {
// label = labels[i];
// max = probabilities[i];
// }
// }
// return new LabelAndProbability(Label, (accuracy/n));
return new LabelAndProbability(Label, votes);
}
public svm_node[] buildingBean2SvmNode(BuldingBean buldingBean) {
// TODO Auto-generated method stub
svm_node p0 = new svm_node();
p0.index = 0;
p0.value = buldingBean.getBaseElectric();
svm_node p1 = new svm_node();
p1.index = 1;
p1.value = buldingBean.getAirElectric();
svm_node[] p = {p0, p1};
return p;
}
