-
Notifications
You must be signed in to change notification settings - Fork 0
/
Copy pathanalysis_replication.Rmd
184 lines (139 loc) · 5.21 KB
/
analysis_replication.Rmd
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
---
title: "Analysis replication v3"
author: "Léopold Guyot"
date: "`r Sys.Date()`"
output: html_document
---
```{r setup, include=FALSE}
knitr::opts_chunk$set(echo = TRUE)
```
# Packages
```{r}
library(gbutils)
library(scpdata)
library(scp)
library(tidyverse)
library(ggthemes)
```
# Functions
```{r}
mse_compute <- function(matrix, reference){
result <- mean((scores(reference)[,1] - scores(matrix)[,1])^2+(scores(reference)[,2] - scores(matrix)[,2])^2)
return(result)
}
retrieve_totss <- function(kmn_obj){
if(isNA(kmn_obj)|is.null(kmn_obj$tot.withinss)){
return(NA)
}
return(kmn_obj$tot.withinss)
}
retrieve_betss <- function(kmn_obj){
if(isNA(kmn_obj) | is.null(kmn_obj$tot.withinss)){
return(NA)
}
return(kmn_obj$betweenss)
}
accuracy <- function(kmn, set){
conf_table <- table(colData(leduc)[unlist(attributes(kmn$cluster)), "SampleType"], kmn$cluster)
accs <- c((conf_table[1,1]+conf_table[2,2])/sum(conf_table), (conf_table[1,2]+conf_table[2,1])/sum(conf_table))
return(max(accs))
}
retrieve_acc <- function(kmn_obj,ref){
if(isNA(kmn_obj) | is.null(kmn_obj) ){
return(NA)
}
acc <- accuracy(kmn_obj, ref)
return(acc)
}
```
# Loading of data
```{r}
kmn_na <- readRDS(file = "data_output/na_kmn_v3.rds")
kmn_na_random <- readRDS(file ="data_output/na_kmn_random_v3.rds")
kmn_svd <- readRDS(file = "data_output/svd_kmn_v3.rds")
svd_comp_knn <- readRDS(file = "data_output/svd_comp_kmn_v3.rds")
complete_kmn <- readRDS(file = "data_output/comp_pca_kmn_v3.rds")
leduc <- leduc2022()
```
# betss, totss and acc
```{r}
kmn_to_df <- function(kmn_list, set){
kmn_totss <- lapply(kmn_list, function(x){
lapply(x, function(y){
lapply(y, retrieve_totss)
})
})
kmn_totss
kmn_betss <- lapply(kmn_list, function(x){
lapply(x, function(y){
lapply(y, retrieve_betss)
})
})
kmn_acc <- lapply(kmn_list, function(x){ #fix
lapply(x, function(y){
lapply(y, retrieve_acc, leduc)
})
})
rates = c(0, 0.1, 0.2, 0.3, 0.4, 0.5, 0.6, 0.7, 0.8, 0.9, 0.95)
df <- data.frame("method" =rep(c("nipals", "ppca"), 88),"na_rate"=rep(sort(rep(rates, 2)), 8), "totss"=unlist(kmn_totss),"betss"=unlist(kmn_betss), "acc" = unlist(kmn_acc))
return(df)
}
```
# In a dataframe
```{r}
df_complete <- kmn_to_df(complete_kmn, leduc)
df_kmn_no_random <- kmn_to_df(kmn_na, leduc)
df_kmn_random <- kmn_to_df(kmn_na_random, leduc)
mean_svd <- mean(unlist(lapply(kmn_svd, retrieve_acc, leduc)))
mean_svd
```
# Non random generation
```{r}
df_no_random <- df_kmn_no_random %>%
mutate("factor_na" = as.factor(na_rate)) %>%
mutate("bet_tot" = betss/totss) %>%
mutate("acc_per" = acc*100)
df_no_random_mean <- df_no_random %>%
group_by(na_rate, method) %>%
summarize("mean_totss" = mean(totss), "mean_betss" = mean(betss), "mean_bet_tot" = mean(bet_tot), "mean_acc" = mean(acc), "median_acc" = median(acc))
ggplot(df_no_random, aes(x = factor_na, y = bet_tot, color = method))+geom_boxplot()+facet_wrap(vars(method))
ggplot(df_no_random, aes(x = factor_na, y = acc_per, color = method))+geom_boxplot()+facet_grid(rows = vars(method))+
geom_hline(aes(yintercept = mean_svd*100), linewidth = 0.4, color="darkgreen", linetype="dashed")+
ylab("Accuracy (%)")+xlab("missing value rates")+theme(legend.key.height = unit(1, "cm"),
legend.key.width = unit(1,"cm"),
legend.text = element_text(size=10),
legend.title = element_text(size = 10))
ggsave(filename = "figures/replication_3_no_random_acc.pdf", width = 9, height = 8)
```
# Random generation
```{r}
df_random <- df_kmn_random %>%
mutate("factor_na" = as.factor(na_rate)) %>%
mutate("bet_tot" = betss/totss)
kmn_results_mean <- df_random %>%
group_by(na_rate, method) %>%
summarize("mean_totss" = mean(totss), "mean_betss" = mean(betss), "mean_bet_tot" = mean(bet_tot), "mean_acc" = mean(acc), "median_acc" = median(acc))
ggplot(df_random, aes(x = factor_na, y = bet_tot, color = method))+geom_boxplot()+facet_wrap(vars(method))
ggplot(df_random, aes(x = factor_na, y = acc, color = method))+geom_boxplot()+facet_grid(rows = vars(method))+ylab("Accuracy")+xlab("NA rates")+
theme(legend.key.height = unit(1, "cm"),
legend.key.width = unit(1,"cm"),
legend.text = element_text(size=13),
legend.title = element_text(size = 13))
?theme
ggsave(filename = "figures/replication_3_random_acc.pdf")
```
# Complete subset
```{r}
df_comp <- df_complete %>%
mutate("factor_na" = as.factor(na_rate)) %>%
mutate("bet_tot" = betss/totss)
kmn_results_mean <- df_comp %>%
group_by(na_rate, method) %>%
summarize("mean_totss" = mean(totss), "mean_betss" = mean(betss), "mean_bet_tot" = mean(bet_tot), "mean_acc" = mean(acc), "median_acc" = median(acc))
ggplot(df_comp, aes(x = factor_na, y = bet_tot, color = method))+geom_boxplot()+facet_wrap(vars(method))
ggplot(df_comp, aes(x = factor_na, y = acc, color = method))+geom_boxplot()+facet_grid(rows = vars(method))+ylab("Accuracy")+xlab("NA rates")+theme(legend.key.height = unit(1, "cm"),
legend.key.width = unit(1,"cm"),
legend.text = element_text(size=13),
legend.title = element_text(size = 13))
ggsave(filename = "figures/replication_3_comp_acc.png", width = 12, height = 9.5 )
```