cluster init refactoring, fix best sort, add km_score()

1 files changed, 51 insertions, 0 deletions

diff --git a/km-init-forgy.c b/km-init-forgy.c
new file mode 100644
index 0000000..687c528
--- /dev/null
+++ b/km-init-forgy.c
@@ -0,0 +1,51 @@
+#include <stdbool.h> // bool
+#include <string.h> // memset()
+#include "util.h"
+#include "km.h"
+
+// init a set with num_clusters clusters of shape num_floats by picking
+// random initial points from the set
+bool
+km_init_forgy(
+  km_set_t * const cs,
+  const size_t num_clusters,
+  const km_set_t * const set,
+  km_rand_t * const rs
+) {
+  const size_t num_floats = set->shape.num_floats,
+               stride = sizeof(float) * num_floats;
+
+  // init cluster shape
+  const km_shape_t shape = {
+    .num_floats = num_floats,
+    .num_ints = 1,
+  };
+
+  // get random row offsets
+  size_t rows[num_clusters];
+  if (!km_rand_fill_sizes(rs, num_clusters, rows)) {
+    // return failure
+    return false;
+  }
+
+  // generate random cluster centers
+  float floats[num_floats * num_clusters];
+  for (size_t i = 0; i < num_clusters; i++) {
+    const size_t row_num = rows[i] % set->num_rows;
+    const float * const row_floats = km_set_get_row(set, row_num);
+    memcpy(floats + i * num_floats, row_floats, stride);
+  }
+
+  // FIXME: should probably be heap-allocated
+  int ints[num_clusters];
+  memset(ints, 0, sizeof(ints));
+
+  // init cluster set
+  if (!km_set_init(cs, &shape, num_clusters)) {
+    // return failure
+    return false;
+  }
+
+  // add data, return result
+  return km_set_push(cs, num_clusters, floats, ints);
+}