From df06fe6084b1c1673348050309a3a0cb99b6634f Mon Sep 17 00:00:00 2001
From: Paul Duncan <pabs@pablotron.org>
Date: Sun, 3 Feb 2019 15:10:44 -0500
Subject: add cluster support to gen-data.rb, show centroids on generated png

---
 km-find.c | 23 +++++++++++++++++++++++
 1 file changed, 23 insertions(+)

(limited to 'km-find.c')

diff --git a/km-find.c b/km-find.c
index 2dd1ba1..7ed9a6a 100644
--- a/km-find.c
+++ b/km-find.c
@@ -5,6 +5,14 @@
 
 #define MIN_CLUSTER_DISTANCE 0.0001
 
+static float
+get_score(
+  const float mean_distance,
+  const size_t num_empty
+) {
+  return 1.0 / (mean_distance + num_empty);
+}
+
 typedef struct {
   float distance_sum,
         variance_sum;
@@ -95,6 +103,7 @@ km_find(
     return false;
   }
 
+  float best_score = 0.0;
   for (size_t i = 2; i < cbs->max_clusters; i++) {
     for (size_t j = 0; j < cbs->num_tests; j++) {
       // init cluster set
@@ -130,6 +139,20 @@ km_find(
       // emit result
       cbs->on_data(&result, cb_data);
 
+      // score result
+      float score = get_score(result.mean_distance, result.num_empty_clusters);
+
+      if (score > best_score) {
+        // emit new best result
+        if (cbs->on_best && !cbs->on_best(score, &cs, cb_data)) {
+          // return failure
+          return false;
+        }
+
+        // update best score
+        best_score = score;
+      }
+
       // finalize cluster set
       if (!cbs->on_fini(&cs, cb_data)) {
         // return failure
-- 
cgit v1.2.3