-
Notifications
You must be signed in to change notification settings - Fork 2
/
Copy pathEvaluation_of_scenarios_rho.R
208 lines (174 loc) · 8.74 KB
/
Evaluation_of_scenarios_rho.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
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
#' # Evaluation of the scenarios
#'
#+ warning = FALSE, message = FALSE, error = FALSE, include = TRUE, echo = FALSE
# Packages
library(RColorBrewer)
library(ggplot2)
library(plyr)
#'
#' The truth is the same within each scenario.
#'
#' The default parameters are as follows:
#'
#' - simulation is conducted on a grid of 300*300
#' - environmental covariate coefficient of 1.2
#' - scale parameter kappa for matern covariance of 0.05
#' - variance parameter sigma2x of matern covariance of 2
#' - mean log intensity of point process of -1
#' - 150 structured samples
#' - probability of sampling strata rep(c(0.5, 0.3, 0.1, 0.05, 0.01),5)
#' - qsize of 1
#'
#'
#' ## Structured sample size scenario
#'
#'
#+ warning = FALSE, message = FALSE, error = FALSE, include = TRUE, echo = FALSE
# set up code and parameters for summaries
source('parallel_summary.R')
n_runs = 500
n_by = 4
n_tot = n_runs*n_by
files <- list.files(path = ".", pattern = "Rho")
files <- files[-c(1:2)]
# create a summary of all runs of this scenario
summary_scenario_rho <- as.data.frame(t(mapply(summary_wrapper, files,
MoreArgs = list(
summary = "summary", n_tot,
n_by), SIMPLIFY = T))) # transposed to look clearer
raw_scenario_rho <- mapply(summary_wrapper, files,
MoreArgs = list(summary = "raw", n_tot,
n_by), SIMPLIFY = F)
# summary table
row.names(summary_scenario_rho) <- str_sub(row.names(summary_scenario_rho), 5, -7)
# add new column of the number of samples
# need to remove the model name - can be tricky as different lengths
scenario_names <- unlist(row.names(summary_scenario_rho))
# model names need to be in set order so remove completely
model_names = c("unstructuredcov", "unstructured", "structured", "jointtwo", "jointcov", "joint")
# easiest in loop
for(i in 1:length(model_names)){
scenario_names <- str_replace(scenario_names, model_names[i], "")
}
summary_scenario_rho$Scenario <- as.numeric(scenario_names)
summary_scenario_rho[,1:9] <- unlist(summary_scenario_rho[,1:9]) # need to unlist to save
write.csv(summary_scenario_rho, "SummaryTable_rho.csv", row.names=T)
#' ### Table
#'
#+ warning = FALSE, message = FALSE, error = FALSE, include = TRUE, echo = FALSE
summary_scenario_rho
#'
#' ### Figures
#+ warning = FALSE, message = FALSE, error = FALSE, include = TRUE, echo = FALSE
# join all of the correlation estimates into a dataframe so can use ggplot
# do this from the raw data
plotting_data <- summary_plot_function(raw_scenario_rho, scenario = "rho_", n_runs, type="summary")
# relevel model column
plotting_data$model <- factor(plotting_data$model, level = c("structured",
"unstructured",
"joint",
"unstructuredcov",
"jointcov", "jointtwo"))
plotting_data$model <- revalue(plotting_data$model, c("unstructured" = "PO only (B)",
"unstructuredcov" = "PO with \nbias \ncovariate (D)",
"structured" = "PA only (A)",
"joint" = "IDM (C)",
"jointcov" = "IDM with \nbias \ncovariate (E)",
"jointtwo" = "IDM with \nsecond spatial \nfield (F)"))
plotting_data$scenario <- as.numeric(plotting_data$scenario)
plotting_data$scenario[which(plotting_data$scenario==9)] <- plotting_data$scenario[which(plotting_data$scenario==9)]/10
plotting_data$scenario[which(plotting_data$scenario==95)] <- plotting_data$scenario[which(plotting_data$scenario==95)]/100
plotting_data$scenario[which(plotting_data$scenario==99)] <- plotting_data$scenario[which(plotting_data$scenario==99)]/100
# now plot
# set manual colours
manual_colours <- c("darkblue", "grey50")
# Plot at least 95% of the estimates for each scenario
y_correlation <- round(y_limits(plotting_data, "correlation"),2)
Correlation <- ggplot(plotting_data, aes(as.factor(scenario), correlation))+
scale_fill_manual(values=manual_colours, name = "",
labels = c("Structured only",
"Unstructured only",
"Joint",
"Unstructured with \nbias \ncovariate",
"Joint with \nbias \ncovariate",
"Joint with \nsecond spatial field"))+
geom_boxplot(aes(fill=as.factor(model)), outlier.shape=NA)+
theme_classic()+
theme(legend.position = "none")+
xlab("Rho (correlation between actual bias and covariate describing bias)")+
ylab("Correlation between prediction and truth")+
facet_wrap(~as.factor(model), nrow=1, scales="free_x")+
theme(axis.text.x = element_text(angle = 0, hjust = 1))+
ylim(c(y_correlation))
Correlation
ggsave(filename = "CorrelationPlot_rho.png", plot=last_plot(),
width = 20, height = 10, units="cm", dpi=300)
y_env <- round(y_limits(plotting_data, "env"),2)
Environment <- ggplot(plotting_data, aes(as.factor(scenario), env))+
scale_fill_manual(values=manual_colours, name = "",
labels = c("Structured only",
"Unstructured only",
"Joint",
"Unstructured with \nbias \ncovariate",
"Joint with \nbias \ncovariate",
"Joint with \nsecond spatial field"))+
geom_boxplot(aes(fill=as.factor(model)), outlier.shape=NA)+
geom_hline(aes(yintercept = 2), linetype="dashed", color = "red")+
theme_classic()+
theme(legend.position = "none")+
xlab("Rho (correlation between actual bias and covariate describing bias)")+
ylab("Environmental covariate estimate")+
ylim(c(0,6))+
facet_wrap(~as.factor(model), nrow=1, scales="free_x")+
theme(axis.text.x = element_text(angle = 0, hjust = 1))
Environment
ggsave(filename = "EnvironmentPlot_rho.png", plot=last_plot(),
width = 20, height = 10, units="cm", dpi=300)
# add an extra plot of width of credible intervals
y_width <- round(y_limits(plotting_data, "width"),2)
Environment_CI <- ggplot(plotting_data, aes(as.factor(scenario), width))+
scale_fill_manual(values=manual_colours, name = "",
labels = c("Structured only",
"Unstructured only",
"Joint",
"Unstructured with \nbias \ncovariate",
"Joint with \nbias \ncovariate",
"Joint with \nsecond spatial field"))+
geom_boxplot(aes(fill=as.factor(model)), outlier.shape=NA)+
theme_classic()+
theme(legend.position = "none")+
xlab("Rho (correlation between actual bias and covariate describing bias)")+
ylab("Width of credible interval for environmental covariate")+
ylim(c(0,35))+
facet_wrap(~as.factor(model), nrow=1, scales="free_x")+
theme(axis.text.x = element_text(angle = 0, hjust = 1))
Environment_CI
ggsave(filename = "EnvironmentPlotCI_Rho.png", plot=last_plot(),
width = 20, height = 10, units="cm", dpi=300)
y_mae <- round(y_limits(plotting_data, "mae"),2)
MAE <- ggplot(plotting_data, aes(as.factor(scenario), mae))+
scale_fill_manual(values=manual_colours, name = "",
labels = c("Structured only",
"Unstructured only",
"Joint",
"Unstructured with \nbias \ncovariate",
"Joint with \nbias \ncovariate",
"Joint with \nsecond spatial field"))+
geom_boxplot(aes(fill=as.factor(model)), outlier.shape=NA)+
theme_classic()+
theme(legend.position = "none")+
xlab("Rho (correlation between actual bias and covariate describing bias)")+
ylab("MAE")+
ylim(c(0,1.5))+
facet_wrap(~as.factor(model), nrow=1, scales="free_x")+
theme(axis.text.x = element_text(angle = 0, hjust = 1))
MAE
ggsave(filename = "MAEPlot_Rho.png", plot=last_plot(),
width = 20, height = 10, units="cm", dpi=300)
#' ## Table of proportion of env estimate in CI
#'
#+ warning = FALSE, message = FALSE, error = FALSE, include = TRUE, echo = FALSE
# calculate the proportion of simulations where true environmental beta
# in credibility interval
prop_env_in_CI <- summary_plot_function(raw_scenario_rho, scenario = "rho_", n_runs, type="CI")
cbind(row.names(summary_scenario_rho),prop_env_in_CI)