-
Notifications
You must be signed in to change notification settings - Fork 9
/
PracticalAssignment_3_Amelia_Mice.R
196 lines (183 loc) · 10.2 KB
/
PracticalAssignment_3_Amelia_Mice.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
# Practical assignment 4
# Dr Reza Rafiee, Oct. 2020
# Analysis: multiple imputation modelling using Amelia and mice packages in R
# The input csv file includes numeric values with some missing data
# The aim is (but not limited) to the following items:
# 1) Read the CSV file including 220 samples with 17 features,
# 2) Assess the fraction of missing across all features and samples,
# 3) Use multiple imputation techniques to impute the missingness (both Amelia and mice packages),
# 4) Visualise the imputation results (including imputed values vs. observed, etc.) ,
# 5) Finally comparing the obtained results from two techniques using a scatter plot.
#-----------------------------------------------------------------------------------------
#----------------------------------------------------------------------------------------------------
# Libraries will be in this section
#
library(Amelia)
library(mice)
#----------------------------------------------------------------------------------------------------
#~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
# User functions will be in this section
#~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
# Addition of the two input matrices element-wise
add_element_wise <- function(x1_matrix,y1_matrix) {
Sum_matrix <- x1_matrix # the sum of two matrix will be here
for (i in 1:nrow(x1_matrix)) { # i is the index of row
for (j in 1:ncol(x1_matrix)) {
Sum_matrix[i,j] <- Sum_matrix[i,j] + y1_matrix[i,j]
} # for j
} # for i
return(Sum_matrix) # return the sum of two matrices
} # function add_element_wise
#~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
#~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
# Chagning the range of input matrix into [0 1]
Data_Range_Into_01 <- function(x){(x-min(x,na.rm=T))/(max(x,na.rm=T)-min(x,na.rm=T))}
#~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
# Imputation using mice package
#~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
MultipleImputationModeling <- function(Input_Sample)
{
# Columns must be features and rows must be samples
# Show the missing data pattern
md.pattern(Input_Sample)
# Column should be variables (features or gene)
# maxit: a scalar giving the number of iterations. The default is 5
# m: Number of multiple imputations.
NumofIteration <- 5
imputated_by_mice <- mice(Input_Sample,m=20,maxit=NumofIteration,seed = 123)
#summary(imputated_by_mice)
### blue is observed, red is imputed
#stripplot(imputated_by_mice, PLG_S5, pch = 19, xlab = "Imputation number") #In general, we would like the imputations to be plausible, i.e., values that could have been observed if they had not been missing.
#stripplot(imputated_by_mice, MMP3_S2, pch = 19, xlab = "Imputation number") #In general, we would like the imputations to be plausible, i.e., values that could have been observed if they had not been missing.
#stripplot(imputated_by_mice, RGS8_S5, pch = 19, xlab = "Imputation number") #In general, we would like the imputations to be plausible, i.e., values that could have been observed if they had not been missing.
#imputated_by_mice2 <- mice.mids(imputated_by_mice)
# verification
#identical(imputated_by_mice$imp, imputated_by_mice2$imp)
### density plot of head circumference per imputation
### blue is observed, red is imputed
#densityplot(imputated_by_mice, ~PLG_S5|.imp)
#densityplot(imputated_by_mice, ~MMP3_S2|.imp)
#densityplot(imputated_by_mice, ~RGS8_S5|.imp)
#xyplot(imputated_by_mice,...)
#==========================================
X <- complete(imputated_by_mice, action = "long", include = TRUE)[, -2]
test <- as.mids(X, where = NULL, .imp = ".imp", .id = ".id")
is.mids(test)
Test_dat <- complete(test, action = "long", include = TRUE)
#================================================
#output1 <- matrix(unlist(imputated_by_mice2), ncol = 17, byrow = TRUE)
#imputated_by_mice$data[1:17]
#firstSample <- imputated_by_mice$data[3,1:17]
ColumnNoPlus2MoreVaraibles <- ncol(Input_Sample) + 2
Imputed_mat <- matrix(nrow = nrow(Input_Sample) ,ncol = ColumnNoPlus2MoreVaraibles ,0.0) # ncol = 19
nrow(Test_dat)
#for (impx in 0:5) # number of iterations
for(idx in 1:nrow(Input_Sample)) # number of samples
{
for (j in 3:ColumnNoPlus2MoreVaraibles) # number of columns
{
if (is.na(Test_dat[idx,j]))
{
sum.imp <- 0
for (impx in 1:NumofIteration+1) # number of iterations: 6
{
sum.imp <- Test_dat[nrow(Input_Sample)*impx+idx,j] + sum.imp
}
Imputed_mat[idx,j] <- sum.imp/(NumofIteration+1)
}
else
{
Imputed_mat[idx,j] <- Test_dat[idx,j]
}
}
}
Imputed_mat <- Imputed_mat[,-c(1:2)]
Imputed_mat <- t(Imputed_mat)
colnames(Imputed_mat) <- rownames(Input_Sample)
rownames(Imputed_mat) <- colnames(Input_Sample)
return(Imputed_mat)
}
#~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
#----------------------------------------------------------------------------------------------------
#~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
#####################################################################################################
# Main programme - Start
#####################################################################################################
# Set working directory and reading the CSV file including 220 samples with 17 features.
getwd()
setwd("D:/Live")
# Read my input csv file from working directory
Input_dataset <- as.data.frame(t(read.csv("PA3_CSC3062_With_Missing_RR_2019_220.csv",row.names = 1)))
class(Input_dataset) # [1] "data.frame"
attributes(Input_dataset)
dim(Input_dataset) # [1] 220 17
#----------------------------------------------------------------------------------------------------
# First analysis: assess the fraction of missing across all features and samples
#----------------------------------------------------------------------------------------------------
# Running multiple imputation using Amelia package (m=20)
Number_of_imputation <- 5
datasets_models_using_amelia <- amelia(x = Input_dataset, m = Number_of_imputation, p2s = 1, frontend = FALSE)
summary(datasets_models_using_amelia)
# Fraction Missing
# feature_1 0.009090909
# feature_2 0.027272727
# feature_3 0.000000000
# feature_4 0.018181818
# feature_5 0.013636364
# feature_6 0.013636364
# feature_7 0.018181818
# feature_8 0.027272727
# feature_9 0.031818182
# feature_10 0.013636364
# feature_11 0.022727273
# feature_12 0.018181818
# feature_13 0.027272727
# feature_14 0.013636364
# feature_15 0.009090909
# feature_16 0.009090909
# feature_17 0.009090909
########################################################################################
# Adding element wise of the complete datasets obtained from amelia
Sum_matrix <- t(datasets_models_using_amelia$imputations[[1]]) # The first complete dataset
for (i_index in 2:Number_of_imputation) {
Sum_matrix <- add_element_wise(Sum_matrix,t(datasets_models_using_amelia$imputations[[i_index]]))
} # for
avg_matrix_elementwise <- Sum_matrix/Number_of_imputation
min(avg_matrix_elementwise) # what is the min of element in the final matrix: -0.1179939
max(avg_matrix_elementwise) # what is the max of element in the final matrix: 1.061036
#~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
# or using the following for loop without a function
Sum_matrix_2 <- t(datasets_models_using_amelia$imputations[[1]]) # The first complete dataset
for (i_index in 2:Number_of_imputation) {
Sum_matrix_2 <- Sum_matrix_2 + t(datasets_models_using_amelia$imputations[[i_index]])
} # for
# Find the average
avg_matrix_elementwis_2 <- Sum_matrix_2/Number_of_imputation
Final_imputed_dataset_using_amelia <- as.data.frame(t(avg_matrix_elementwis_2))
min(avg_matrix_elementwis_2) # what is the min of element in the final matrix: -0.1179939
max(avg_matrix_elementwise) # what is the max of element in the final matrix: 1.061036
########################################################################################
# Imputation using mice package
NumofIteration <- 5
Number_of_imputation <- 20
imputated_by_mice <- mice(Input_dataset,m=Number_of_imputation, maxit=NumofIteration,seed = 12345)
Sum_matrix_3 <- complete(imputated_by_mice,1) # The first imputed dataset obtained from mice package
# Adding
for (i in 2:Number_of_imputation) {
Sum_matrix_3 <- Sum_matrix_3 + complete(imputated_by_mice,i)
}
Sum_matrix_3 <- Sum_matrix_3/Number_of_imputation
Final_imputed_dataset_using_mice <- as.data.frame(Sum_matrix_3)
#|||||||||||||||||||||||||||||||||||||||||
# or you could call the follwoing function
# Call the function for multiple imputation using mice
Final_imputed_dataset_using_mice <- MultipleImputationModeling(Input_dataset) # Numeric dataset
Final_imputed_dataset_using_mice <- as.data.frame(t(Final_imputed_dataset_using_mice)) # as dataframe
# Till here, we have imputed two datasets obtained from Amelia and mice
# Now, try to compare the results
matplot(Final_imputed_dataset_using_mice, Final_imputed_dataset_using_amelia, type="p")
########################################################################################
########################################################################################
####################################################################################################
# Main programme - END
#####################################################################################################