From 6908f7be4c7e620ac1ee4db9bdbf4cc5100e9e5a Mon Sep 17 00:00:00 2001
From: Joel Klinghed <the_jk@spawned.biz>
Date: Thu, 18 Nov 2021 00:22:10 +0100
Subject: travel: Improve trip location estimation by removing outliers

Outlier value (> 1.43 Z-score) based on what was needed for a trip
I took long ago. Will have to fiddle with the value I assume.
---
 src/travel.cc | 74 +++++++++++++++++++++++++++++++++++++++++++++++------------
 1 file changed, 59 insertions(+), 15 deletions(-)

(limited to 'src')

diff --git a/src/travel.cc b/src/travel.cc
index f8adf50..57d226a 100644
--- a/src/travel.cc
+++ b/src/travel.cc
@@ -13,6 +13,8 @@
 #include "weak_ptr.hh"
 
 #include <algorithm>
+#include <map>
+#include <math.h>
 #include <mutex>
 #include <string>
 #include <utility>
@@ -454,32 +456,74 @@ private:
     call_when_loaded_.clear();
   }
 
+  static Location get_mean_location(std::vector<Location> locations) {
+    if (locations.empty())
+      return Location();
+
+    auto it = locations.begin();
+    Location mean = *it;
+    for (++it; it != locations.end(); ++it) {
+      mean.lat += it->lat;
+      mean.lng += it->lng;
+    }
+    mean.lat /= locations.size();
+    mean.lng /= locations.size();
+    return mean;
+  }
+
+  static Location get_center_location(std::vector<Location> locations) {
+    while (true) {
+      if (locations.empty())
+        return Location();
+
+      Location mean = get_mean_location(locations);
+
+      float variance = 0;
+      std::vector<float> location_z;
+      for (auto const& loc : locations) {
+        float deviation_square = std::pow(loc.lat - mean.lat, 2.f) +
+          std::pow(loc.lng - mean.lng, 2.f);
+        variance += deviation_square;
+        location_z.push_back(std::sqrt(deviation_square));
+      }
+      float standard_deviation = std::sqrt(variance / locations.size());
+
+      for (auto& z : location_z)
+        z /= standard_deviation;
+
+      // Remove outliers (if any)
+      bool removed_any = false;
+      size_t i = 0;
+      while (i < locations.size()) {
+        if (std::abs(location_z[i]) > 1.43f) {
+          locations.erase(locations.begin() + i);
+          location_z.erase(location_z.begin() + i);
+          removed_any = true;
+        } else {
+          ++i;
+        }
+      }
+
+      if (!removed_any)
+        return mean;
+    }
+  }
+
   void cleanup_trip(TripImpl& trip_impl) {
     // Sort by date
     trip_impl.sort_media();
 
     trip_impl.setup_days();
 
-    // TODO: Remove outliers
-    // TODO: Use some weighted median instead of average
-    Location loc;
-    size_t count = 0;
+    std::vector<Location> locations;
     for (size_t i = 0; i < trip_impl.media_count(); ++i) {
       auto& media = trip_impl.media(i);
       if (media.location().empty())
         continue;
-      if (loc.empty()) {
-        loc = media.location();
-        count = 1;
-      } else {
-        loc.lat += media.location().lat;
-        loc.lng += media.location().lng;
-        ++count;
-      }
+      locations.push_back(media.location());
     }
-    loc.lat /= count;
-    loc.lng /= count;
-    trip_impl.set_location(loc);
+
+    trip_impl.set_location(get_center_location(std::move(locations)));
   }
 
   static std::optional<ImageInfo> get_image_info(std::string id,
-- 
cgit v1.2.3-70-g09d2