void kmedoids::pam(const dissimilarity_matrix& distance, size_t k, const object_id *initial_medoids) {
if (k > distance.size1()) {
throw std::logic_error("Attempt to run PAM with more clusters than data.");
}
if (distance.size1() != distance.size2()) {
throw std::logic_error("Error: distance matrix is not square!");
}
// first get this the right size.
cluster_ids.resize(distance.size1());
// size cluster_ids appropriately and randomly pick initial medoids
if (initial_medoids) {
medoid_ids.clear();
copy(initial_medoids, initial_medoids + k, back_inserter(medoid_ids));
} else {
init_medoids(k, distance);
}
// set tolerance equal to epsilon times mean magnitude of distances.
// Note that distances *should* all be non-negative.
double tolerance = epsilon * sum(distance) / (distance.size1() * distance.size2());
while (true) {
// initial cluster setup
total_dissimilarity = assign_objects_to_clusters(matrix_distance(distance));
//vars to keep track of minimum
double minTotalCost = DBL_MAX;
medoid_id minMedoid = 0;
object_id minObject = 0;
//iterate over each medoid
for (medoid_id i=0; i < k; i++) {
//iterate over all non-medoid objects
for (object_id h = 0; h < cluster_ids.size(); h++) {
if (is_medoid(h)) continue;
//see if the total cost of swapping i & h was less than min
double curCost = cost(i, h, distance);
if (curCost < minTotalCost) {
minTotalCost = curCost;
minMedoid = i;
minObject = h;
}
}
}
// bail if we can't gain anything more (we've converged)
if (minTotalCost >= -tolerance) break;
// install the new medoid if we found a beneficial swap
medoid_ids[minMedoid] = minObject;
cluster_ids[minObject] = minMedoid;
}
if (sort_medoids) sort();
}
// Main loop for an instance of the algorithm.
double run() {
size_t i;
size_t j;
size_t k;
printf("Initialization.\n");
init_medoids();
if(verbose) print_medoids(medoids);
for(k = 0; k < clustc; ++k) {
for(j = 0; j < dmatrixc; ++j) {
weights[k][j] = 1.0;
}
}
if(verbose) print_weights(weights);
update_memb();
if(verbose) print_memb(memb);
double prev_adeq = 0.0;
double adeq = adequacy_obj(false);
printf("Adequacy: %.20lf\n", adeq);
double diff = fabs(adeq - prev_adeq);
for(i = 1; i <= max_iter && diff > epsilon; ++i) {
printf("Iteration %d.\n", i);
prev_adeq = adeq;
adequacy_cluster(false);
update_medoids();
adeq = adequacy_cluster(true);
if(verbose) {
print_medoids(medoids);
printf("Adequacy1: %.20lf\n", adeq);
}
adequacy_cluster(false);
update_weights();
adeq = adequacy_cluster(true);
if(verbose) {
print_weights(weights);
printf("Adequacy2: %.20lf\n", adeq);
}
adequacy_obj(false);
update_memb();
adeq = adequacy_obj(true);
if(verbose) print_memb(memb);
printf("Adequacy: %.20lf\n", adeq);
if(dgt(adeq, prev_adeq)) {
printf("Warn: current adequacy is greater than "
"previous iteration (%.20lf)\n",
adeq - prev_adeq);
}
diff = fabs(adeq - prev_adeq);
}
printf("Adequacy difference threshold reached (%.20lf).\n",
diff);
return adeq;
}