exp_class_stagedtrees/plot_simulations.R at main · stagedtrees/exp_class_stagedtrees · GitHub

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
library(ggplot2)
library(reshape2)

classifiers <- c("st_bhc_3db",
                 "st_bhc_5db",
                 "st_bhc_tan_cl",
                 "st_bhc_tan_hc",
                 "st_kmeans_cmi",
                 "st_hclust_cmi",
                 "bnc_3db",
                 "bnc_5db",
                 "bnc_tan_hc",
                 "bnc_tan_cl")

colors <- colorspace::qualitative_hcl(n = 11, palette = "dark3")
names(colors) <- classifiers

TABLE <- readRDS("TABLE_SIMULATION.rds")
AVG <- apply(TABLE, 1:4, mean, na.rm = TRUE)
SD <- apply(TABLE, 1:4, sd, na.rm = TRUE)
#AVG['time',,,] <- log(AVG['time',,,], base = 10)
dd <- melt(AVG,value.name = "value", na.rm = TRUE)
Q1 <- apply(TABLE, 1:4, quantile, probs = 0.05, na.rm = TRUE)
Q2 <- apply(TABLE, 1:4, quantile, probs = 0.95, na.rm = TRUE)
dd$q1 <- melt(Q1,value.name = "q1", na.rm = TRUE)$q1
dd$q2 <- melt(Q2,value.name = "q2", na.rm = TRUE)$q2
dd$gen = "random_sevt"

TABLE <- readRDS("TABLE_SIMULATION_LINEAR.rds")
AVG <- apply(TABLE, 1:4, mean, na.rm = TRUE)
ddl <- melt(AVG,value.name = "value", na.rm = TRUE)
Q1 <- apply(TABLE, 1:4, quantile, probs = 0.05, na.rm = TRUE)
Q2 <- apply(TABLE, 1:4, quantile, probs = 0.95, na.rm = TRUE)
ddl$q1 <- melt(Q1,value.name = "q1", na.rm = TRUE)$q1
ddl$q2 <- melt(Q2,value.name = "q2", na.rm = TRUE)$q2
ddl$gen = "linear"

TABLE <- readRDS("TABLE_SIMULATION_XOR.rds")
AVG <- apply(TABLE, 1:4, mean, na.rm = TRUE)
ddx <- melt(AVG,value.name = "value", na.rm = TRUE)
Q1 <- apply(TABLE, 1:4, quantile, probs = 0.05, na.rm = TRUE)
Q2 <- apply(TABLE, 1:4, quantile, probs = 0.95, na.rm = TRUE)
ddx$q1 <- melt(Q1,value.name = "q1", na.rm = TRUE)$q1
ddx$q2 <- melt(Q2,value.name = "q2", na.rm = TRUE)$q2
ddx$gen = "xor"

alld <- rbind(dd, ddl, ddx)
alld$gen <- factor(alld$gen, levels = c("random_sevt", "linear", "xor"))
#levels(alld$stat)[1] <- "log-time"
alld$stat <- factor(alld$stat, levels = c("accuracy", "f1", "time"))
ggplot(alld[alld$classifier %in% classifiers,]) + geom_line(aes(x = p, y = value, group = classifier, color = classifier)) +
  #geom_ribbon(aes(x = p, ymin = q1, ymax = q2,
  #                group = classifier, fill = classifier), alpha = 0.2, color = NA) +
  facet_grid(rows = vars(stat), cols = vars(gen),  scales = "free") + theme_bw() + ylab("") +
  #scale_color_brewer(type = "qual") +
  scale_color_manual(values = colors) +
  #scale_fill_manual(values = colors) +
  #scale_y_log10() +
  #scale_color_viridis_d()+
  theme(legend.position = "bottom", legend.title = element_blank())

ggsave(filename = "res_simulation.pdf", width = 7, height = 4)