-
Notifications
You must be signed in to change notification settings - Fork 1
/
sourcecode.R
163 lines (145 loc) · 8.46 KB
/
sourcecode.R
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
library(tidyverse)
library(magrittr)
library(DataCombine)
library(car)
setwd("/Users/jiayan/Downloads/bayesianinference_HIVmortality-master/input_files")
data <- read_csv("HIVdf.csv")
#Change data class to avoide error
data$PerCapInc <- as.numeric(data$PerCapInc)
data$Pop <- as.numeric(data$Pop)
# Compute 4 inequality measures by the definition in chap 2.2.3
Theil <- data %$% left_join(., setNames(aggregate(PerCapInc*Pop,
by = list(CountyName = CountyName), sum), c("CountyName","IncCounty")), # aggregate the per-cap income by County
by = c("CountyName")) %>% mutate(IncZCTA = PerCapInc*Pop, zctaCount = 1, # create zctaCount for T4
IncZCTA.CPI = PerCapInc*Pop/CPI2000, cInc=IncZCTA/IncCounty, # Inc.CPI-adjusted/ZCTA & Inc/ZCTA
IncNY = sum(IncZCTA), IncNY.CPI = sum(IncZCTA.CPI), sInc= IncZCTA/IncNY, # total income share in NY, ZCTA income in relation to NY income
sInc.CPI = IncZCTA.CPI/IncNY.CPI) %$% # CPI-adjusted sINC
left_join(., setNames(aggregate(Pop, by = list(CountyName = CountyName), sum), # agrregate population by county
c("CountyName","PopCounty")), by = c("CountyName")) %>%
mutate(cWe = Pop/PopCounty, sWe = Pop/sum(Pop),
t1j = sInc*log(sInc/sWe), t2j = cInc*log(cInc/cWe), #calculate T1 & T2
t3j = sInc.CPI*log(sInc.CPI/sWe), t4j = ifelse(t2j < 0,1,0)) %$% #calculate T3 & partial T4
left_join(setNames(aggregate(zctaCount, by = list(CountyName), sum), c("CountyName", "zctaCount")),
left_join(setNames(aggregate(t3j, by = list(CountyName), sum), c("CountyName", "T3")), # nested join operation for all Theil indexes
left_join(setNames(aggregate(t4j, by = list(CountyName), sum), c("CountyName", "T4")),
left_join(setNames(aggregate(t1j, by = list(CountyName),sum), c("CountyName", "T1")),
setNames(aggregate(t2j, by = list(CountyName),sum), c("CountyName", "T2")),
by = c("CountyName")), by = c("CountyName")), by = c("CountyName")),
by = c("CountyName")) %>% #nested join of T1+T2 -> T4 -> T3
InsertRow(., c("Hamilton",1,0,0,0,0),21) %>% mutate_at(c(-1), as.numeric) %>% #insert Hamilton row, (insert row function coerce cols to char type); convert cols back to numeric
mutate(T4 = T4/zctaCount) %>% select(CountyName, T1,T2,T3,T4) # re-order sequence
# create T1 Boxplot
Boxplot(Theil$T1, id = list(labels = Theil$CountyName, cex = 3, n = 2),
ann = FALSE, cex.axis = 2.5, cex = 3)
title(ylab = parse(text=paste("T", "^1 ","*Index")), cex.lab = 3, line = 4)
# create T2 Boxplot
Boxplot(Theil$T2, id = list(labels = Theil$CountyName, cex = 3),
ann = FALSE, cex.axis = 2.5, cex = 3)
title(ylab = parse(text = paste("T^2 ", "*Index")), cex.lab = 3, line = 4)
# create T3 Boxplot
Boxplot(Theil$T3, id = list(labels = Theil$CountyName, cex = 3, n = 2),
ann = FALSE, cex.axis = 2.5, cex = 3)
title(ylab = parse(text = paste("T^3 *Index")), cex.lab = 3, line = 4)
# create T4 Boxplot
Boxplot(Theil$T4, id = list(labels = Theil$CountyName, cex = 3),
ann = FALSE, cex.axis = 2.5, cex = 3)
title(ylab = parse(text = paste("T^4 *Index")), cex.lab = 3, line = 4)
# Retrieve Poverty Info
pov <- data %>% select(CountyName, PovertyPercentage) %>% unique() %$%
InsertRow(.,c("Hamilton", mean(PovertyPercentage)),21) %>%
mutate_at(c(2), as.numeric)
pov$PovertyPercentage <- round(pov$PovertyPercentage,2)
# Retrieve Unemployment Info
une <- data %>% select(CountyName, InLaborForce, Unemployed) %$%
left_join(setNames(aggregate(InLaborForce, FUN = sum, by = list(CountyName)),
c("CountyName", "LaborForceSum")), setNames(aggregate(Unemployed, FUN = sum,
by = list(CountyName)), c("CountyName", "UnemployedSum")),
by = c("CountyName")) %>%
mutate(UnemploymentRate = UnemployedSum/LaborForceSum) %$%
InsertRow(., c("Hamilton",0,0,mean(UnemploymentRate)),21) %>%
mutate_at(c(2,3,4), as.numeric)
# Create Poverty Boxplot
par(mar = c(1,7,1,1))
Boxplot(pov$PovertyPercentage, id = list(labels = pov$CountyName, cex = 3), ann = FALSE, cex.axis = 2.5, cex = 3)
title(ylab = "Poverty Rate (%)", cex.lab = 3, line = 4.5)
#create Unemployment Boxplot
Boxplot(une$UnemploymentRate, id = list(labels = une$CountyName, cex = 3), ann = FALSE, cex.axis = 2.5, cex = 3)
title(ylab = "Unemployment Rate (%)", cex.lab = 3, line = 4.5)
# MCMC diagnostics
library(coda)
library(ggmcmc)
library(mcmcplots)
setwd("./input_files/coda")
file.List <- list.files(".")
file.List <- file.List[sort.list(file.List)]
# load coda files and generate MCMC objects for each models
for(i in seq(25, length(file.List), 4)){
subname <- sub("CODA.*","",file.List[i]) # get the substring before "CODA"
chain1 <- read.coda(output.file = paste0(subname, "CODA_chain1.txt"), index.file = paste0(subname, "CODA_index.txt"))
chain2 <- read.coda(output.file = paste0(subname, "CODA_chain2.txt"), index.file = paste0(subname, "CODA_index.txt"))
chain3 <- read.coda(output.file = paste0(subname, "CODA_chain3.txt"), index.file = paste0(subname, "CODA_index.txt"))
var.name <- paste(gsub("[_]","", sub("CODA.*","",file.List[i])), sep = "")
assign(var.name, as.mcmc.list(list(chain1, chain2, chain3))) # create variable dynamically
assign(paste0(var.name, ".gg"), ggs(get(var.name))) #retrieve the variable object as input to ggs function
}
glo.var <- names(as.list(.GlobalEnv)) %>% .[grep(".gg", .)] %>%
str_remove_all(., ".gg") # get all variable name that is not with .gg
# to generate all html files with trace, density, autocorrelation plot in one page
for(i in 1:length(glo.var)){
dir.create(paste0("/Users/jiayan/Downloads/HTML_Diagnostics/",glo.var[i])) #create subfolder to store all html files
mcmcplot(get(glo.var[i]), style = "plain", greek = T,
dir = paste0("/Users/jiayan/Downloads/HTML_Diagnostics/",glo.var[i]),
filename = glo.var[i], extension = "html", heading = glo.var[i])
}
# Gelman & Rubin diagnostics
# CAUTION: high computing power, run function without loops if necessary
for(i in 1:length(glo.var)){
gelman.diag(get(glo.var[i]))
gelman.plot(get(glo.var[i]))
}
# to generate boxplot for each parameters using the chains value retrieved from coda files
dc.M4EinterT4E <- M4EinterT4E.gg %>% filter(Parameter %in% c("dc[1]", "dc[2]", "dc[3]",
"dc[4]", "dc[5]")) %>% select(3,4) %>% mutate_at(c(1), as.character) %>%
mutate(param = case_when(Parameter == "dc[1]" ~ "2000",
Parameter == "dc[2]" ~ "2001", Parameter == "dc[3]" ~ "2002",
Parameter == "dc[4]" ~ "2003", Parameter == "dc[5]" ~ "2004")) %>%
select(param, value)
par(mar = c(4,5,1,1))
# create M4EinterT4E temporal boxplot
boxplot(value ~ param, dc.M4EinterT4E, ylab = expression(paste("c"[t])),
xlab = "", cex.lab = 2, cex.axis = 2)
dc.M2inter <- M2inter.gg %>% filter(Parameter %in% c("dc[1]", "dc[2]", "dc[3]",
"dc[4]", "dc[5]")) %>% select(3,4) %>% mutate_at(c(1), as.character) %>%
mutate(param = case_when(Parameter == "dc[1]" ~ "2000",
Parameter == "dc[2]" ~ "2001", Parameter == "dc[3]" ~ "2002",
Parameter == "dc[4]" ~ "2003", Parameter == "dc[5]" ~ "2004")) %>%
select(param, value)
# create M2inter temporal boxplot
boxplot(value ~ param, dc.m2inter, ylab = expression(paste("c"[t])), xlab = "",
cex.lab = 2, cex.axis = 2)
# to generate boxplot for spatial precision prior, tau
tau.M4EinterT4E <- M4EinterT4E.gg %>% filter(Parameter %in% c("tau[1]",
"tau[2]", "tau[3]", "tau[4]", "tau[5]")) %>% select(3,4) %>%
mutate_at(c(1), as.character) %>%
mutate(param = case_when(Parameter == "tau[1]" ~ "tau^1",
Parameter == "tau[2]" ~ "tau^2", Parameter == "tau[3]" ~ "tau^3",
Parameter == "tau[4]" ~ "tau^4", Parameter == "tau[5]" ~ "tau^5")) %>%
select(param, value)
# create M4EinterT4E spatial precision boxplot
boxplot(value ~ param, tau.M4EinterT4E, ylab = "values", xlab = "",
cex.lab = 2, cex.axis = 2, xaxt = "n")
axis(1, at=1:5, labels=c(parse(text=paste("tau^(1)")),
parse(text=paste("tau^(2)")),parse(text=paste("tau^(3)")),
parse(text=paste("tau^(4)")),parse(text=paste("tau^(5)"))), cex.axis = 2)
tau.M2inter <- M2inter.gg %>% filter(Parameter %in% c("tau[1]", "tau[2]",
"tau[3]", "tau[4]", "tau[5]")) %>% select(3,4) %>% mutate_at(c(1),
as.character) %>% mutate(param = case_when(Parameter == "tau[1]" ~ "tau^1",
Parameter == "tau[2]" ~ "tau^2", Parameter == "tau[3]" ~ "tau^3",
Parameter == "tau[4]" ~ "tau^4", Parameter == "tau[5]" ~ "tau^5")) %>%
select(param, value)
# create M2inter spatial precision boxplot
boxplot(value ~ param, tau.M2inter, ylab = "values", xlab = "", cex.lab = 2,
cex.axis = 2, xaxt = "n")
axis(1, at=1:5, labels=c(parse(text=paste("tau^(1)")),
parse(text=paste("tau^(2)")),parse(text=paste("tau^(3)")),
parse(text=paste("tau^(4)")),parse(text=paste("tau^(5)"))), cex.axis = 2)