From 375f29ab4306821c888fd9ef0637f9ab2879e375 Mon Sep 17 00:00:00 2001
From: Ben Green <ben@SEASITs-MacBook-Pro.local>
Date: Thu, 2 Jul 2015 00:41:49 -0400
Subject: in which the model works!!

---
 R Scripts/create-hyper-lcc.R      | 126 --------------------------------------
 R Scripts/generate-hyper-lcc.R    | 126 ++++++++++++++++++++++++++++++++++++++
 R Scripts/predict-victims-plots.R |   3 +-
 3 files changed, 128 insertions(+), 127 deletions(-)
 delete mode 100644 R Scripts/create-hyper-lcc.R
 create mode 100644 R Scripts/generate-hyper-lcc.R

(limited to 'R Scripts')

diff --git a/R Scripts/create-hyper-lcc.R b/R Scripts/create-hyper-lcc.R
deleted file mode 100644
index 786b694..0000000
--- a/R Scripts/create-hyper-lcc.R	
+++ /dev/null
@@ -1,126 +0,0 @@
-library(igraph)
-setwd("~/Documents/Cascade Project/")
-load('Raw Data/lcc.RData')
-
-lcc_verts = get.data.frame(lcc,'vertices')
-lcc_edges = get.data.frame(lcc,'edges')
-
-
-##### Create new vertices dataframe
-cols = c('id','ir_no','sex','race','dob','age.arrest','arrest.day',
-         'gang.member','gang.name','faction.name')
-hyp_lcc_verts = data.frame(matrix(ncol=length(cols)+1,nrow=0))
-colnames(hyp_lcc_verts) = c(cols,'spawn.date')
-  
-ptm = proc.time()
-ri = 1
-for (i in 1:dim(lcc_verts)[1]){
-  if (i%%10000==0) print(i)
-  if (lcc_verts$vic[i]){
-    if (lcc_verts$vic.nonfatal[i]>0){
-      # create nodes for each nonfatal shooting
-      for (nf in 1:lcc_verts$vic.nonfatal[i]){
-        hyp_lcc_verts[ri,cols] = lcc_verts[i,cols]
-        hyp_lcc_verts$id[ri] = ri
-        hyp_lcc_verts$vic[ri] = T
-        hyp_lcc_verts$vic.type[ri] = 'nonfatal'
-        if(nf==1){
-          hyp_lcc_verts$vic.day[ri] = lcc_verts$nonfatal_day_1[i]
-          hyp_lcc_verts$spawn.date[ri] = 0
-        } else if(nf==2){
-          hyp_lcc_verts$vic.day[ri] = lcc_verts$nonfatal_day_2[i]
-          hyp_lcc_verts$spawn.date[ri] = lcc_verts$nonfatal_day_1[i]
-        } else if(nf==3){
-          hyp_lcc_verts$vic.day[ri] = lcc_verts$nonfatal_day_3[i]
-          hyp_lcc_verts$spawn.date[ri] = lcc_verts$nonfatal_day_2[i]
-        } else if(nf==4){
-          hyp_lcc_verts$vic.day[ri] = lcc_verts$nonfatal_day_4[i]
-          hyp_lcc_verts$spawn.date[ri] = lcc_verts$nonfatal_day_3[i]
-        } else if(nf==5){
-          hyp_lcc_verts$vic.day[ri] = lcc_verts$nonfatal_day_5[i]
-          hyp_lcc_verts$spawn.date[ri] = lcc_verts$nonfatal_day_4[i]
-        }
-        ri = ri+1
-      }
-      # if no fatal infection, create uninfected duplicate
-      if (!lcc_verts$vic.fatal[i]){
-        hyp_lcc_verts[ri,cols] = lcc_verts[i,cols]
-        hyp_lcc_verts$id[ri] = ri
-        hyp_lcc_verts$vic[ri] = F
-        hyp_lcc_verts$vic.type[ri] = NA
-        hyp_lcc_verts$vic.day[ri] = NA
-        hyp_lcc_verts$spawn.date[ri] = max(lcc_verts$nonfatal_day_1[i],lcc_verts$nonfatal_day_2[i],
-                                           lcc_verts$nonfatal_day_3[i],lcc_verts$nonfatal_day_4[i],
-                                           lcc_verts$nonfatal_day_5[i],na.rm=T)
-        ri = ri+1
-      }
-    }
-    # create a node for each fatal shooting
-    # if also nonfatal shootings, spawn at last nonfatal shooting
-    if (lcc_verts$vic.fatal[i]){
-      hyp_lcc_verts[ri,cols] = lcc_verts[i,cols]
-      hyp_lcc_verts$id[ri] = ri
-      hyp_lcc_verts$vic[ri] = T
-      hyp_lcc_verts$vic.type[ri] = 'fatal'
-      hyp_lcc_verts$vic.day[ri] = lcc_verts$fatal_day[i] 
-      if (lcc_verts$vic.nonfatal[i]>0){
-        hyp_lcc_verts$spawn.date[ri] = max(lcc_verts$nonfatal_day_1[i],lcc_verts$nonfatal_day_2[i],
-                                           lcc_verts$nonfatal_day_3[i],lcc_verts$nonfatal_day_4[i],
-                                           lcc_verts$nonfatal_day_5[i],na.rm=T)
-      } else {
-        hyp_lcc_verts$spawn.date[ri] = 0
-      }
-      ri = ri+1
-    }
-  } 
-  # create an uninfected node for each uninfected person
-  else{
-    hyp_lcc_verts[ri,cols] = lcc_verts[i,cols]
-    hyp_lcc_verts$id[ri] = ri
-    hyp_lcc_verts$vic[ri] = F
-    hyp_lcc_verts$vic.type[ri] = NA
-    hyp_lcc_verts$vic.day[ri] = NA
-    hyp_lcc_verts$spawn.date[ri] = 0
-    ri = ri+1
-  }
-}
-print(proc.time()-ptm) # 1.5 hrs
-row.names(hyp_lcc_verts) = NULL
-n.nodes = sum(sum(V(lcc)$vic.fatal),sum(V(lcc)$vic.nonfatal),
-              sum(V(lcc)$vic.nonfatal>0 & !V(lcc)$vic.fatal),sum(V(lcc)$vic==FALSE))
-stopifnot(dim(hyp_lcc_verts)[1] == n.nodes)
-
-##### Create new edgelist
-print('Edges')
-hyp_lcc_edges = data.frame(from=0, to=0, weight=0)
-ei = 1
-ptm = proc.time()
-for(i in 1:dim(hyp_lcc_verts)[1]){
-  if (i%%10000==0) print(i)
-  ego_id = hyp_lcc_verts$id[i]
-  ego_ir = hyp_lcc_verts$ir_no[i]
-  alter_irs = union(lcc_edges$from[which(lcc_edges$to==ego_ir)],
-                    lcc_edges$to[which(lcc_edges$from==ego_ir)])
-#   alter_irs = union(ego_ir, alter_irs) # add edges to other infected selves
-  alter_ids = hyp_lcc_verts$id[which(hyp_lcc_verts$ir_no %in% alter_irs)]
-  alter_ids = alter_ids[ego_id<alter_ids] # avoid double-counting edges
-  for(alter_id in alter_ids){
-    weight=Inf
-    alter_ir = hyp_lcc_verts$ir_no[alter_id]
-    if(ego_ir!=alter_ir){
-      edge_id = which((lcc_edges$from %in% c(ego_ir,alter_ir) + lcc_edges$to %in% c(ego_ir,alter_ir))==2)
-      weight = lcc_edges$weight[edge_id]
-    }
-    hyp_lcc_edges[ei,] = c(ego_id, alter_id, weight)
-    ei = ei + 1
-  }
-}
-print(proc.time()-ptm) #10 hrs
-
-##### Create new graph
-hyp_lcc = graph.data.frame(hyp_lcc_edges, directed=FALSE, vertices=hyp_lcc_verts)
-
-save(hyp_lcc_edges, hyp_lcc_verts, hyp_lcc, file='Results/hyper-lcc.RData')
-write.csv(hyp_lcc_edges, file='Results/hyp_lcc_edges.csv')
-write.csv(hyp_lcc_verts[,c('id','spawn.date','vic','vic.type','vic.day')], file='Results/hyp_lcc_verts.csv')
-
diff --git a/R Scripts/generate-hyper-lcc.R b/R Scripts/generate-hyper-lcc.R
new file mode 100644
index 0000000..786b694
--- /dev/null
+++ b/R Scripts/generate-hyper-lcc.R	
@@ -0,0 +1,126 @@
+library(igraph)
+setwd("~/Documents/Cascade Project/")
+load('Raw Data/lcc.RData')
+
+lcc_verts = get.data.frame(lcc,'vertices')
+lcc_edges = get.data.frame(lcc,'edges')
+
+
+##### Create new vertices dataframe
+cols = c('id','ir_no','sex','race','dob','age.arrest','arrest.day',
+         'gang.member','gang.name','faction.name')
+hyp_lcc_verts = data.frame(matrix(ncol=length(cols)+1,nrow=0))
+colnames(hyp_lcc_verts) = c(cols,'spawn.date')
+  
+ptm = proc.time()
+ri = 1
+for (i in 1:dim(lcc_verts)[1]){
+  if (i%%10000==0) print(i)
+  if (lcc_verts$vic[i]){
+    if (lcc_verts$vic.nonfatal[i]>0){
+      # create nodes for each nonfatal shooting
+      for (nf in 1:lcc_verts$vic.nonfatal[i]){
+        hyp_lcc_verts[ri,cols] = lcc_verts[i,cols]
+        hyp_lcc_verts$id[ri] = ri
+        hyp_lcc_verts$vic[ri] = T
+        hyp_lcc_verts$vic.type[ri] = 'nonfatal'
+        if(nf==1){
+          hyp_lcc_verts$vic.day[ri] = lcc_verts$nonfatal_day_1[i]
+          hyp_lcc_verts$spawn.date[ri] = 0
+        } else if(nf==2){
+          hyp_lcc_verts$vic.day[ri] = lcc_verts$nonfatal_day_2[i]
+          hyp_lcc_verts$spawn.date[ri] = lcc_verts$nonfatal_day_1[i]
+        } else if(nf==3){
+          hyp_lcc_verts$vic.day[ri] = lcc_verts$nonfatal_day_3[i]
+          hyp_lcc_verts$spawn.date[ri] = lcc_verts$nonfatal_day_2[i]
+        } else if(nf==4){
+          hyp_lcc_verts$vic.day[ri] = lcc_verts$nonfatal_day_4[i]
+          hyp_lcc_verts$spawn.date[ri] = lcc_verts$nonfatal_day_3[i]
+        } else if(nf==5){
+          hyp_lcc_verts$vic.day[ri] = lcc_verts$nonfatal_day_5[i]
+          hyp_lcc_verts$spawn.date[ri] = lcc_verts$nonfatal_day_4[i]
+        }
+        ri = ri+1
+      }
+      # if no fatal infection, create uninfected duplicate
+      if (!lcc_verts$vic.fatal[i]){
+        hyp_lcc_verts[ri,cols] = lcc_verts[i,cols]
+        hyp_lcc_verts$id[ri] = ri
+        hyp_lcc_verts$vic[ri] = F
+        hyp_lcc_verts$vic.type[ri] = NA
+        hyp_lcc_verts$vic.day[ri] = NA
+        hyp_lcc_verts$spawn.date[ri] = max(lcc_verts$nonfatal_day_1[i],lcc_verts$nonfatal_day_2[i],
+                                           lcc_verts$nonfatal_day_3[i],lcc_verts$nonfatal_day_4[i],
+                                           lcc_verts$nonfatal_day_5[i],na.rm=T)
+        ri = ri+1
+      }
+    }
+    # create a node for each fatal shooting
+    # if also nonfatal shootings, spawn at last nonfatal shooting
+    if (lcc_verts$vic.fatal[i]){
+      hyp_lcc_verts[ri,cols] = lcc_verts[i,cols]
+      hyp_lcc_verts$id[ri] = ri
+      hyp_lcc_verts$vic[ri] = T
+      hyp_lcc_verts$vic.type[ri] = 'fatal'
+      hyp_lcc_verts$vic.day[ri] = lcc_verts$fatal_day[i] 
+      if (lcc_verts$vic.nonfatal[i]>0){
+        hyp_lcc_verts$spawn.date[ri] = max(lcc_verts$nonfatal_day_1[i],lcc_verts$nonfatal_day_2[i],
+                                           lcc_verts$nonfatal_day_3[i],lcc_verts$nonfatal_day_4[i],
+                                           lcc_verts$nonfatal_day_5[i],na.rm=T)
+      } else {
+        hyp_lcc_verts$spawn.date[ri] = 0
+      }
+      ri = ri+1
+    }
+  } 
+  # create an uninfected node for each uninfected person
+  else{
+    hyp_lcc_verts[ri,cols] = lcc_verts[i,cols]
+    hyp_lcc_verts$id[ri] = ri
+    hyp_lcc_verts$vic[ri] = F
+    hyp_lcc_verts$vic.type[ri] = NA
+    hyp_lcc_verts$vic.day[ri] = NA
+    hyp_lcc_verts$spawn.date[ri] = 0
+    ri = ri+1
+  }
+}
+print(proc.time()-ptm) # 1.5 hrs
+row.names(hyp_lcc_verts) = NULL
+n.nodes = sum(sum(V(lcc)$vic.fatal),sum(V(lcc)$vic.nonfatal),
+              sum(V(lcc)$vic.nonfatal>0 & !V(lcc)$vic.fatal),sum(V(lcc)$vic==FALSE))
+stopifnot(dim(hyp_lcc_verts)[1] == n.nodes)
+
+##### Create new edgelist
+print('Edges')
+hyp_lcc_edges = data.frame(from=0, to=0, weight=0)
+ei = 1
+ptm = proc.time()
+for(i in 1:dim(hyp_lcc_verts)[1]){
+  if (i%%10000==0) print(i)
+  ego_id = hyp_lcc_verts$id[i]
+  ego_ir = hyp_lcc_verts$ir_no[i]
+  alter_irs = union(lcc_edges$from[which(lcc_edges$to==ego_ir)],
+                    lcc_edges$to[which(lcc_edges$from==ego_ir)])
+#   alter_irs = union(ego_ir, alter_irs) # add edges to other infected selves
+  alter_ids = hyp_lcc_verts$id[which(hyp_lcc_verts$ir_no %in% alter_irs)]
+  alter_ids = alter_ids[ego_id<alter_ids] # avoid double-counting edges
+  for(alter_id in alter_ids){
+    weight=Inf
+    alter_ir = hyp_lcc_verts$ir_no[alter_id]
+    if(ego_ir!=alter_ir){
+      edge_id = which((lcc_edges$from %in% c(ego_ir,alter_ir) + lcc_edges$to %in% c(ego_ir,alter_ir))==2)
+      weight = lcc_edges$weight[edge_id]
+    }
+    hyp_lcc_edges[ei,] = c(ego_id, alter_id, weight)
+    ei = ei + 1
+  }
+}
+print(proc.time()-ptm) #10 hrs
+
+##### Create new graph
+hyp_lcc = graph.data.frame(hyp_lcc_edges, directed=FALSE, vertices=hyp_lcc_verts)
+
+save(hyp_lcc_edges, hyp_lcc_verts, hyp_lcc, file='Results/hyper-lcc.RData')
+write.csv(hyp_lcc_edges, file='Results/hyp_lcc_edges.csv')
+write.csv(hyp_lcc_verts[,c('id','spawn.date','vic','vic.type','vic.day')], file='Results/hyp_lcc_verts.csv')
+
diff --git a/R Scripts/predict-victims-plots.R b/R Scripts/predict-victims-plots.R
index 2ac62c8..553aa89 100644
--- a/R Scripts/predict-victims-plots.R	
+++ b/R Scripts/predict-victims-plots.R	
@@ -1,5 +1,5 @@
 ##### Plot results
-hist(correct_rank3,150,xlim=c(0,vcount(lcc)),col=rgb(0,0,1,1/8),
+hist(correct_rank1,150,xlim=c(0,vcount(lcc)),col=rgb(0,0,1,1/8),
      xlab='Risk Ranking of Victims',main='')
 hist(correct_rank1,150,xlim=c(0,vcount(lcc)),col=rgb(1,0,1,1/8),add=T)
 hist(correct_rank2,150,xlim=c(0,vcount(lcc)),col=rgb(0,0,1,1/8),add=T)
@@ -36,6 +36,7 @@ legend("topleft", inset=0.05,
        c("Demographics Model", "Cascade Model", "Combined Model"), 
        fill=c(rgb(0,0,1,1/2),rgb(1,0,0,1/2),rgb(0,1,0,1/2)))
 box(which='plot')
+
 par(new=T)
 counts = counts/(100/nvics)
 barplot(counts, 
-- 
cgit v1.2.3-70-g09d2