added plotting

Clustering_by_Kmeans.py

@@ -129,5 +129,40 @@ def plot_clusters(groups, numclusters):
 
     plt.show()
 
+def findSuitablek(data, minimum, maximum):
+    mumcluster = minimum
+    groups = cluster(data, numcluster) 
+    diff = []
+    for i in range(minimum, maximum+1):
+        groups = cluster(data,numcluster)
+        minlen = min([len(i) for i in groups])
+        maxlen = max([len(i) for i in groups])
+        diff.append(maxlen - minlen)
+    mindiff = min(diff)
+    for i in len(diff):
+        if(diff[i] == mindiff):
+            return i+minimum
+
+
+def groupclusters(data, numcluster, seats):
+    groups = cluster(data, numcluster)
+    gcluster = [[] [] []]
+
+    for i in groups:
+        if(len(groups) < 0.6 * seats):
+            gcluster[0].append(i)
+        elif (len(groups) <1.2 * seats):
+            gcluster[1].append(i)
+        else:
+            gcluster[2].append(i)
+
+    for i in gcluster[2]:
+        getgroup = cluster(i, len(i)/(0.85 * seats))
+        for j in getgroup:
+            gcluster[1].append(i)
 
-plot_clusters(cluster(data, 4),4)
+    return gcluster[1]
+
+
+plot_clusters(cluster(data, 4))
+

presentation.txt

@@ -4,23 +4,31 @@ outline:
     b. constraints and their priorities.
     c. impact of this problem in real time, eg. millions of dollars can
         be saved by having efficient soln.
+Garbage Truck Scheduling
+Reduces Carbon footprint
+this problem impacts areas such as these
+
     d. the general methods of solving this; eg. linear programming and
         heuristic approaches in it, ant colonization and why these fail.
 
 (finish above in around 5 mins)
 
 2. Our Solution
+
+    Make a flowchart of the process
+
     a. Use TSP as we want to minimise cost eventually(explain TSP too).
     b. Problem in using TSP though, is its bad time complexity.
     c. Hence use clustering to improve time; explain the derivation
     d. This serves another purpose as well i.e bus size and percent occupancy,
         how: assign 1 bus for each cluster and run TSP for that cluster
     e. Now we have to make sure size of each cluster is >85% of bus cap
         and number of clusters is <= total number of buses available,
-        to handle this, explain the process given by Chaithanya.
+        to handle this, explain the process given by Chaithanya.(60%, 120%)
+
     f. We use KMeans algo for clustering, use the longitude and latitude
         data to make a 2d plot of the bus stops and their group decided
-        by the algo.
+        by the algo.(also explain the number of clusters decided)
     g. To handle time window, run TSP on weighted graph, alpha beta
         method and plot a graph to find alpha such that time is within
         limit.