From c37f720eae9ea5eb2d85794a4822b622daa73e9c Mon Sep 17 00:00:00 2001
From: Schankula Christopher <schankuc@mcmaster.ca>
Date: Mon, 2 Apr 2018 17:23:11 -0400
Subject: [PATCH] first draft of (fast) clustering
---
src/Main.java | 6 +++
src/search/BasicSearch.java | 2 +-
src/search/BasicSearchResult.java | 76 ++++++++++++++++++++++++-------
src/search/CC.java | 2 +
src/search/RecordCluster.java | 5 ++
src/search/RedBlackTree.java | 7 +--
6 files changed, 76 insertions(+), 22 deletions(-)
diff --git a/src/Main.java b/src/Main.java
index 8c91e26..520b924 100644
--- a/src/Main.java
+++ b/src/Main.java
@@ -18,6 +18,7 @@ import search.BST;
import search.BasicSearch;
import search.BasicSearchResult;
import search.Histogram;
+import search.RecordCluster;
import sort.Bound;
import sort.GeneralRange;
import sort.KDT;
@@ -71,6 +72,11 @@ public class Main {
public static void init() {
System.out.println("Welcome!");
+
+ BasicSearchResult bsr = BasicSearch.range(159512, 1960, 2016);
+ ArrayList<RecordCluster> rc = (ArrayList<RecordCluster>) bsr.cluster(1000000);
+ System.out.println(rc.size());
+
while(true) {
System.out.println("Main Menu");
System.out.println("Available commands:");
diff --git a/src/search/BasicSearch.java b/src/search/BasicSearch.java
index 59b8326..9a8cd23 100644
--- a/src/search/BasicSearch.java
+++ b/src/search/BasicSearch.java
@@ -65,6 +65,6 @@ public class BasicSearch {
}
double time = sw.elapsedTime();
- return new BasicSearchResult(results, time);
+ return new BasicSearchResult(taxonId, yearLo, yearHi, results, time);
}
}
diff --git a/src/search/BasicSearchResult.java b/src/search/BasicSearchResult.java
index 7ab7827..9c0ed43 100644
--- a/src/search/BasicSearchResult.java
+++ b/src/search/BasicSearchResult.java
@@ -1,27 +1,28 @@
package search;
import java.util.ArrayList;
+import java.util.HashMap;
import data.Record;
-<<<<<<< HEAD
+import sort.Bound;
import sort.GeneralRange;
-=======
+import sort.RangeHelper;
+
import java.lang.Math;
->>>>>>> branch 'web' of git@gitlab.cas.mcmaster.ca:schankuc/2XB3.git
public class BasicSearchResult {
+ private final Integer taxonId;
+ private final Integer yearLo;
+ private final Integer yearHi;
private ArrayList<Record> results;
private final double time;
private BST<Integer, Integer> histogram;
private Integer sum;
- private
-
- public static void main(String[] args) {
-
- //Graph g = new Graph();
- }
- public BasicSearchResult(ArrayList<Record> results, double time) {
+ public BasicSearchResult(Integer taxonId, Integer yearLo, Integer yearHi, ArrayList<Record> results, double time) {
+ this.taxonId = taxonId;
+ this.yearLo = yearLo;
+ this.yearHi = yearHi;
this.results = results;
this.time = time;
}
@@ -53,19 +54,64 @@ public class BasicSearchResult {
return this.sum;
}
-<<<<<<< HEAD
/**
*
* @param area
*/
public Iterable<RecordCluster> cluster(double area){
+ HashMap<String, Integer> nodeMap = new HashMap<String, Integer>();
+ HashMap<String, Boolean> marked = new HashMap<String, Boolean>();
+
+ int i = 0;
+ for(Record r: results) {
+ nodeMap.put(r.getOccurId(), i);
+ i++;
+ }
+
+ Graph G = new Graph(results.size());
+
+ double dist = Math.sqrt(area);
+
+
+ for (Record r: results) {
+ if (marked.get(r.getOccurId()) != null) continue;
+ double lat = r.getLatitude();
+ double lon = r.getLongitude();
+ BasicSearchResult res = BasicSearch.range(taxonId, yearLo, yearHi, lat - latRange(dist), lat + latRange(dist), lon - lngRange(dist,lat), lon + lngRange(dist,lat));
+ for (Record r1: res.results()) {
+ marked.put(r1.getOccurId(), true);
+ G.addEdge(nodeMap.get(r.getOccurId()), nodeMap.get(r1.getOccurId()));
+ }
+ }
+ System.out.println("Edges found");
+
+ CC cc = new CC(G);
+
+ ArrayList<RecordCluster> clusters = new ArrayList<RecordCluster>();
+ for(int j = 0; j < cc.count(); j++) {
+ clusters.add(new RecordCluster());
+ }
+
+ for (int j = 0; j < G.V(); j++) {
+ Integer component = cc.id(j);
+ clusters.get(component).addRecord(results.get(j));
+ }
+
+ return clusters;
}
- private ArrayList<GeneralRange> ranges(double lat, double lon, double area){
- ArrayList<GeneralRange>()
+ private ArrayList<GeneralRange<Record>> ranges(double lat, double lon, double area){
+ double dist = Math.sqrt(area);
+
+ ArrayList<GeneralRange<Record>> range = new ArrayList<GeneralRange<Record>>();
+
+ GeneralRange<Record> latRange = RangeHelper.latitude(Bound.LOWHIGH, lat - latRange(dist), lat + latRange(dist));
+ GeneralRange<Record> lonRange = RangeHelper.longitude(Bound.LOWHIGH, lon - lngRange(dist,lat), lon + lngRange(dist,lat));
+
+ return range;
}
-=======
+
private double lngRange(double dist, double lat){
return dist/(Math.cos(lat)*222);
}
@@ -73,6 +119,4 @@ public class BasicSearchResult {
private double latRange(double dist){
return (dist/111)/2;
}
-
->>>>>>> branch 'web' of git@gitlab.cas.mcmaster.ca:schankuc/2XB3.git
}
diff --git a/src/search/CC.java b/src/search/CC.java
index 7726bac..637667f 100644
--- a/src/search/CC.java
+++ b/src/search/CC.java
@@ -1,5 +1,7 @@
package search;
+import java.util.ArrayList;
+
public class CC {
private boolean[] marked;
diff --git a/src/search/RecordCluster.java b/src/search/RecordCluster.java
index e896722..6bfe630 100644
--- a/src/search/RecordCluster.java
+++ b/src/search/RecordCluster.java
@@ -9,6 +9,11 @@ public class RecordCluster {
private ArrayList<Record> records;
private Point centroid;
+ public RecordCluster() {
+ records = new ArrayList<Record>();
+ centroid = new Point(0,0);
+ }
+
public void addRecord(Record r) {
records.add(r);
}
diff --git a/src/search/RedBlackTree.java b/src/search/RedBlackTree.java
index 45bfd91..df9a53a 100644
--- a/src/search/RedBlackTree.java
+++ b/src/search/RedBlackTree.java
@@ -1,5 +1,7 @@
package search;
+import sort.GeneralCompare;
+
public class RedBlackTree<Key, Value> {
private Node<Key, Value> root; // Root of the tree
private GeneralCompare<Key> compare;
@@ -39,11 +41,6 @@ public class RedBlackTree<Key, Value> {
h = h.left();
}
}
-<<<<<<< HEAD
-=======
-
-
->>>>>>> 459508d... get method added to redBlackTree
/**
--
GitLab