In Junzheshao5959/ssmimputedemo: SSM Imputation DEMO(Title Case)
knitr::opts_chunk$set(echo = TRUE)
library(ssmimputedemo)
library(knitr)
generate missing data key
generate.missing_index=function(type,param,n,printFlag=T){
# The function is mainly written by Xinru Wang, edited by Xiaoxuan Cai from 03/17/21 to 03/20/21
# The general idea is from the page 63 of book <<Flexible Imputation of Missing Data>> by Stef van Buuren
# This function generate missing index giving three machenisms: MCAR, MAR, MNAR
# type: [MCAR,MAR,MNAR]
# n: the length of original data
# param: must be a list -> param$data must be a data.frame
# (1) MCAR: <p>
# Example: generate.missing_index(type = "MCAR", n=length(data$y), param = list(p=0.5))
# (2) MAR: (generate missing depending on observed covariates)
# <data> of covariates;
# <MAR.type> (choose in ["increasing","tail","middle])
# <coeff> (#coeff=#covariates+1 for "increasing;
# #coeff=#covariates+1 for "tail";
# #coeff=#covariates+1 for "middle";)
# <MMAR.drawplot> = c(T/F,<name of the covariate>)
# Example: generate.missing_index(type = "MAR", n=length(data$y),
# param =list(data = data.frame(y_1=data$y_1),
# MAR.type = "increasing",
# coeff = c(0.2,-8),
# MAR.drawplot = c(TRUE, "y_1"))
# )
# (3) MNAR: generate missing depending on unobserved values of y and c/x
# <data> unmeasured y or c/x or both
# <MNAR.type> (choose in ["increasing","tail","middle])
# <coeff> (#coeff=#covariates+1 for "increasing;
# #coeff=#covariates+1 for "tail";
# #coeff=#covariates+1 for "middle";)
# <MNAR.drawplot> = c(T/F,<name of the covariate>)
# Example: generate.missing_index(type = "MNAR", n=length(data$y),
# param = ist(data = data.frame(y=data$y,x=data$x),
# MNAR.type = "increasing",
# coeff = c(0.5,-1,0),
# MNAR.drawplot = c(TRUE, "y"))
# )
# Pre-checking
# check parameters: check param$type choose %in% [MCAR,MAR,MNAR]
# check param$data is data frame
# check the length of coeff agrees with the chooseing missing machenism
# type=="MCAR", length(list)==1 note: Xinru change to length(list)==1 (p)
# type=="MAR": length(list)==4 (data,coeff,MAR.type,MAR.drawplot)
# type=="MNAR": length(list)==4 (data,coeff,MNAR.type,MNAR.drawplot)
logistic <- function(x) {exp(x)/(1+exp(x)) }
## check type
if(!type %in% c("MCAR","MAR","MNAR")){
stop("The missing type shoud choose from ''MCAR'',''MAR'',''MNAR''!\n")
}
if(!is.list(param)){
stop("The parameter type is wrong!\n")
}
## Check the length of parameter is 1 for MCAR
if(type=="MCAR"){
if(length(param)!=1)
{ stop("The number of parameters for MCAR shoudl be 1!")}
# create missingness indicator
p=param$p
cat("The missing probability is:",p,"\n")
missing=(2:n)[as.numeric(runif(n-1)<p)==1]
if(T){cat(paste(paste("The missing rate for MCAR is", length(missing)/n, sep = " "),"\n",sep=""))}
}
if(type=="MAR"){
# requirement: param$data must be a data frame
# param$MAR.type in c("increasing","tail","middle")
# param$coeff
# param$MAR.drawplot = T/F
if(length(param)!=4){
stop("We require 4 parameters for MAR.")
}
if(!all(names(param) == c("data","MAR.type","coeff","MAR.drawplot"))){
stop("The input parameter must be [data],[MAR.type],[coeff],[MAR.drawplot]")
}
if(!is.data.frame(param$data) ){
stop("The data in param$data must be a data frame.")
}
if(nrow(param$data)!=n){
stop("The length of given dataset for MAR is wrong.\n")
}
if(!param$MAR.type %in% c("increasing","tail","middle")){
stop("The MAR.type must be increasing/tail/middle.")
}
coeff <- param$coeff
dat <- param$data %>% mutate(intercept=1)
# model is logit(p) = coeff[last]*intercept + X\beta
# the number of coeff is one more than the number of variables
if(param$MAR.type=="increasing"){
if(ncol(param$data) != (length(coeff)-1)){
stop("The number of coefficient for MAR.type 'increasing' is wrong!")
}
if(ncol(dat)!=length(coeff)){
stop("The dat is not correct.")
}
## missing probability
## for one variable logistic(coeff[1]*y_1+coeff[2])
p <- logistic(as.matrix(dat) %*% coeff)
missing=(2:n)[rbinom((n-1),size=1,prob=(1-p[-1]))==0]
if(printFlag){
cat("The missing indicators for MAR 'increasing' are generated!\n")
}
}
# model: logit(p)= coeff[1] + | coeff[2]y_1 (+coeffi[3] c) + coeff[last] |
# the number in coeffi is 2 more than the number of variables
if(param$MAR.type=="tail"){
if(ncol(param$data) != (length(coeff)-2)){
stop("The number of coefficients for MAR.type 'tail' is wrong.")
}
if(ncol(dat)!=length(coeff)-1){
stop("The dat is not correct.")
}
## missing probability
## for one variable logistic(coeff[1]+abs(coeff[2]*y_1+coeff[3]))
p <- logistic(coeff[1]+abs(as.matrix(dat) %*% coeff[-1]))
missing=(2:n)[rbinom((n-1),size=1,prob=(1-p[-1]))==0]
if(printFlag){
cat("The missing indicators for MAR 'tail' are generated!\n")
}
}
# model: logit(p)= coeff[1] - | coeff[2]y_1 (+coeffi[3] c) + coeff[last] |
# the number in coeffi is 2 more than the number of variables
if(param$MAR.type=="middle"){
if(ncol(param$data) != (length(coeff)-2)){
stop("The number of coefficients for MAR.type 'middle' is wrong.")
}
if(ncol(dat)!=length(coeff)-1){
stop("The dat is not correct.")
}
## missing probability
## for one variable logistic(coeff[1]+abs(coeff[2]*y_1+coeff[3]))
p <- logistic(coeff[1]-abs(as.matrix(dat) %*% coeff[-1]))
missing=(2:n)[rbinom((n-1),size=1,prob=(1-p[-1]))==0]
if(printFlag){
cat("The missing indicators for MAR 'middle' are generated!\n")
}
}
if(printFlag){
if(param$MAR.drawplot[1]){
# hist(p)
for(j in 2:length(param$MAR.drawplot)){
data_plot=data.frame(x=dat[-1,param$MAR.drawplot[j]],p=p[-1])
g1 = ggplot(data_plot, aes(x=x, y=p))+
geom_line()+
theme_bw()+
ylab("Missing rate")+
theme(plot.margin = unit(c(1,1,1,1), "cm"),
axis.title.x = element_text(size=15,vjust = -1),
axis.title.y = element_text(size=15,vjust=4),
axis.text = element_text(size=13))+
xlab(param$MAR.drawplot[j])
print(g1)
}
}
}
cat(paste(paste("The missing rate for MAR is", length(missing)/n, sep = " "),"\n",sep=""))
}
if(type=="MNAR"){
# requirement: param$data must be a data frame
# param$MAR.type in c("increasing","tail","middle")
# param$coeff
# param$MAR.drawplot = T/F
if(length(param)!=4){
stop("We require 4 parameters for MAR.")
}
if(!all(names(param) == c("data","MNAR.type","coeff","MNAR.drawplot"))){
stop("The input parameter must be [data],[MNAR.type],[coeff],[MNAR.drawplot]")
}
if(!is.data.frame(param$data) ){
stop("The data in param$data must be a data frame.")
}
if(nrow(param$data)!=n){
stop("The length of given dataset for MAR is wrong.\n")
}
if(!param$MNAR.type %in% c("increasing","tail","middle")){
stop("The MAR.type must be increasing/tail/middle.")
}
coeff <- param$coeff
dat <- param$data %>% mutate(intercept=1)
if(param$MNAR.type=="increasing"){
if(ncol(param$data) != (length(coeff)-1)){
stop("The number of coefficient for MNAR.type 'increasing' is wrong!")
}
if(ncol(dat)!=length(coeff)){
stop("The dat is not correct.")
}
## missing probability
## for one variable logistic(coeff[1]*y_1+coeff[2])
p <- logistic(as.matrix(dat) %*% coeff)
missing=(2:n)[rbinom((n-1),size=1,prob=(1-p[-1]))==0]
if(printFlag){cat("The missing indicators for MNAR 'increasing' are generated!\n")}
}
if(param$MNAR.type=="tail"){
if(ncol(param$data) != (length(coeff)-2)){
stop("The number of coefficients for MNAR.type 'tail' is wrong.")
}
if(ncol(dat)!=length(coeff)-1){
stop("The dat is not correct.")
}
## missing probability
## for one variable logistic(coeff[1]+abs(coeff[2]*y_1+coeff[3]))
p <- logistic(coeff[1]+abs(as.matrix(dat) %*% coeff[-1]))
missing=(2:n)[rbinom((n-1),size=1,prob=(1-p[-1]))==0]
if(printFlag){cat("The missing indicators for MNAR 'tail' are generated!\n")}
}
if(param$MNAR.type=="middle"){
if(ncol(param$data) != (length(coeff)-2)){
stop("The number of coefficients for MNAR.type 'middle' is wrong.")
}
if(ncol(dat)!=length(coeff)-1){
stop("The dat is not correct.")
}
## missing probability
## for one variable logistic(coeff[1]+abs(coeff[2]*y_1+coeff[3]))
p <- logistic(coeff[1]-abs(as.matrix(dat) %*% coeff[-1]))
missing=(2:n)[rbinom((n-1),size=1,prob=(1-p[-1]))==0]
if(printFlag){cat("The missing indicators for MNAR 'middle' are generated!\n")}
}
if(printFlag){
if(param$MNAR.drawplot[1])
{
# hist(p)
for(m in 2:length(param$MNAR.drawplot)){
data_plot=data.frame(x=dat[-1,param$MNAR.drawplot[m]],p=p[-1])
g1 = ggplot(data_plot, aes(x=x, y=p))+
geom_line()+
theme_bw()+
ylab("Missing rates")+
theme(plot.margin = unit(c(1,1,1,1), "cm"),
axis.title.x = element_text(size=15,vjust = -1),
axis.title.y = element_text(size=15,vjust=4),
axis.text = element_text(size=13))+
xlab(param$MNAR.drawplot[m])
print(g1)
}
}
}
cat(paste(paste("The missing rate for MNAR is", length(missing)/n, sep = " "),"\n",sep=""))
}
return(list(missing_index=missing,missing_rates=p))
}
ssmimputedemo::data_stationary
data = data_stationary
data$y_1 = c(NA,data$y[1:999])
data$x_1 = c(NA,data$x[1:999])
data$c_1 = c(NA,data$c[1:999])
index_MCAR = generate.missing_index(type = "MCAR",
n=length(data$y),
param = list(p=0.2))$missing_index
index_MAR = generate.missing_index(type = "MAR",
n=length(data$y),
param =list(data = data.frame(y_1=data$y_1),
MAR.type = "increasing",
coeff = c(0.2,-8),
MAR.drawplot = c(TRUE, "y_1"))
)$missing_index
index_MNAR = generate.missing_index(type = "MNAR",
n=length(data$y),
param = list(data = data.frame(y=data$y,x=data$x),
MNAR.type = "increasing",
coeff = c(0.5,-1,0),
MNAR.drawplot = c(TRUE, "y"))
)$missing_index
data$y[index_MAR] = NA
data$y_1 = c(NA,data$y[1:999])
data$x_1 = c(NA,data$x[1:999])
data$c_1 = c(NA,data$c[1:999])
data = data[c(2:1000),]
data_space_SSMimpute = data
kable(head(data_space_SSMimpute))
imputeTS::ggplot_na_distribution(data_stationary$y, color_missing = "white",color_missing_border = "white", alpha_missing = 0.1)
imputeTS::ggplot_na_distribution(data_space_SSMimpute$y, color_missing = "white",color_missing_border = "white", alpha_missing = 0.1)
printFlag=F
formula="y~y_1+x+x_1+c"
formula_var=unlist(strsplit(unlist(strsplit(formula,"~"))[2],"+",fixed=T))
ss_param=list(inits=c(log(0.25),log(1)),m0=c(40,0.5,-1,-0.5,-1),C0=diag(rep(10^3),5), AR1_coeffi=NULL,rw_coeffi="intercept", v_cp_param=NULL, w_cp_param=NULL,max_iteration=100)
head(data_space_SSMimpute)
result_statespace_SSMimpute1=run.SSMimpute_unanimous_cpts(data_ss_ori=data_space_SSMimpute,formula_var,ss_param_temp=ss_param,
initial_imputation_option="StructTS",
estimate_convergence_cri=0.01,
lik_convergence_cri=0.01,
stepsize_for_newpart=1/3,
max_iteration=100,
cpt_learning_param=list(cpt_method="mean",burnin=1/10,mergeband=20,convergence_cri=10),
cpt_initial_guess_option="ignore",
dlm_option="smooth",m=5,seed=1,printFlag=F)
#kable(result_statespace_SSMimpute1$result_convergence)
#kable(result_statespace_SSMimpute1$result_convergence_mp)
kable(result_statespace_SSMimpute1$result_convergence_mp_addV)
result_statespace_SSMimpute1$estimated_cpts
data_na = result_statespace_SSMimpute1$data_temp
length(data_na$y_1)
data_temp = result_statespace_SSMimpute1$data_temp
missing_part=which(is.na(data_temp$y))[which(is.na(data_temp$y))<nrow(data_temp)]
data_temp$y_1[missing_part+1]=result_statespace_SSMimpute1$y_final
imputeTS::ggplot_na_distribution(data_stationary$y[1:999], color_missing = "pink",color_missing_border = "pink", alpha_missing = 0.9)
imputeTS::ggplot_na_imputations(x_with_na = data_space_SSMimpute$y_1, x_with_imputations = data_na$y_1,x_with_truth = data_stationary$y[1:999])
Junzheshao5959/ssmimputedemo documentation built on Aug. 27, 2022, 8:49 a.m.
R Package Documentation
Browse R Packages
We want your feedback!
Note that we can't provide technical support on individual packages. You should contact the package authors for that.
knitr::opts_chunk$set(echo = TRUE) library(ssmimputedemo) library(knitr)
generate missing data key
generate.missing_index=function(type,param,n,printFlag=T){ # The function is mainly written by Xinru Wang, edited by Xiaoxuan Cai from 03/17/21 to 03/20/21 # The general idea is from the page 63 of book <<Flexible Imputation of Missing Data>> by Stef van Buuren # This function generate missing index giving three machenisms: MCAR, MAR, MNAR # type: [MCAR,MAR,MNAR] # n: the length of original data # param: must be a list -> param$data must be a data.frame # (1) MCAR: <p> # Example: generate.missing_index(type = "MCAR", n=length(data$y), param = list(p=0.5)) # (2) MAR: (generate missing depending on observed covariates) # <data> of covariates; # <MAR.type> (choose in ["increasing","tail","middle]) # <coeff> (#coeff=#covariates+1 for "increasing; # #coeff=#covariates+1 for "tail"; # #coeff=#covariates+1 for "middle";) # <MMAR.drawplot> = c(T/F,<name of the covariate>) # Example: generate.missing_index(type = "MAR", n=length(data$y), # param =list(data = data.frame(y_1=data$y_1), # MAR.type = "increasing", # coeff = c(0.2,-8), # MAR.drawplot = c(TRUE, "y_1")) # ) # (3) MNAR: generate missing depending on unobserved values of y and c/x # <data> unmeasured y or c/x or both # <MNAR.type> (choose in ["increasing","tail","middle]) # <coeff> (#coeff=#covariates+1 for "increasing; # #coeff=#covariates+1 for "tail"; # #coeff=#covariates+1 for "middle";) # <MNAR.drawplot> = c(T/F,<name of the covariate>) # Example: generate.missing_index(type = "MNAR", n=length(data$y), # param = ist(data = data.frame(y=data$y,x=data$x), # MNAR.type = "increasing", # coeff = c(0.5,-1,0), # MNAR.drawplot = c(TRUE, "y")) # ) # Pre-checking # check parameters: check param$type choose %in% [MCAR,MAR,MNAR] # check param$data is data frame # check the length of coeff agrees with the chooseing missing machenism # type=="MCAR", length(list)==1 note: Xinru change to length(list)==1 (p) # type=="MAR": length(list)==4 (data,coeff,MAR.type,MAR.drawplot) # type=="MNAR": length(list)==4 (data,coeff,MNAR.type,MNAR.drawplot) logistic <- function(x) {exp(x)/(1+exp(x)) } ## check type if(!type %in% c("MCAR","MAR","MNAR")){ stop("The missing type shoud choose from ''MCAR'',''MAR'',''MNAR''!\n") } if(!is.list(param)){ stop("The parameter type is wrong!\n") } ## Check the length of parameter is 1 for MCAR if(type=="MCAR"){ if(length(param)!=1) { stop("The number of parameters for MCAR shoudl be 1!")} # create missingness indicator p=param$p cat("The missing probability is:",p,"\n") missing=(2:n)[as.numeric(runif(n-1)<p)==1] if(T){cat(paste(paste("The missing rate for MCAR is", length(missing)/n, sep = " "),"\n",sep=""))} } if(type=="MAR"){ # requirement: param$data must be a data frame # param$MAR.type in c("increasing","tail","middle") # param$coeff # param$MAR.drawplot = T/F if(length(param)!=4){ stop("We require 4 parameters for MAR.") } if(!all(names(param) == c("data","MAR.type","coeff","MAR.drawplot"))){ stop("The input parameter must be [data],[MAR.type],[coeff],[MAR.drawplot]") } if(!is.data.frame(param$data) ){ stop("The data in param$data must be a data frame.") } if(nrow(param$data)!=n){ stop("The length of given dataset for MAR is wrong.\n") } if(!param$MAR.type %in% c("increasing","tail","middle")){ stop("The MAR.type must be increasing/tail/middle.") } coeff <- param$coeff dat <- param$data %>% mutate(intercept=1) # model is logit(p) = coeff[last]*intercept + X\beta # the number of coeff is one more than the number of variables if(param$MAR.type=="increasing"){ if(ncol(param$data) != (length(coeff)-1)){ stop("The number of coefficient for MAR.type 'increasing' is wrong!") } if(ncol(dat)!=length(coeff)){ stop("The dat is not correct.") } ## missing probability ## for one variable logistic(coeff[1]*y_1+coeff[2]) p <- logistic(as.matrix(dat) %*% coeff) missing=(2:n)[rbinom((n-1),size=1,prob=(1-p[-1]))==0] if(printFlag){ cat("The missing indicators for MAR 'increasing' are generated!\n") } } # model: logit(p)= coeff[1] + | coeff[2]y_1 (+coeffi[3] c) + coeff[last] | # the number in coeffi is 2 more than the number of variables if(param$MAR.type=="tail"){ if(ncol(param$data) != (length(coeff)-2)){ stop("The number of coefficients for MAR.type 'tail' is wrong.") } if(ncol(dat)!=length(coeff)-1){ stop("The dat is not correct.") } ## missing probability ## for one variable logistic(coeff[1]+abs(coeff[2]*y_1+coeff[3])) p <- logistic(coeff[1]+abs(as.matrix(dat) %*% coeff[-1])) missing=(2:n)[rbinom((n-1),size=1,prob=(1-p[-1]))==0] if(printFlag){ cat("The missing indicators for MAR 'tail' are generated!\n") } } # model: logit(p)= coeff[1] - | coeff[2]y_1 (+coeffi[3] c) + coeff[last] | # the number in coeffi is 2 more than the number of variables if(param$MAR.type=="middle"){ if(ncol(param$data) != (length(coeff)-2)){ stop("The number of coefficients for MAR.type 'middle' is wrong.") } if(ncol(dat)!=length(coeff)-1){ stop("The dat is not correct.") } ## missing probability ## for one variable logistic(coeff[1]+abs(coeff[2]*y_1+coeff[3])) p <- logistic(coeff[1]-abs(as.matrix(dat) %*% coeff[-1])) missing=(2:n)[rbinom((n-1),size=1,prob=(1-p[-1]))==0] if(printFlag){ cat("The missing indicators for MAR 'middle' are generated!\n") } } if(printFlag){ if(param$MAR.drawplot[1]){ # hist(p) for(j in 2:length(param$MAR.drawplot)){ data_plot=data.frame(x=dat[-1,param$MAR.drawplot[j]],p=p[-1]) g1 = ggplot(data_plot, aes(x=x, y=p))+ geom_line()+ theme_bw()+ ylab("Missing rate")+ theme(plot.margin = unit(c(1,1,1,1), "cm"), axis.title.x = element_text(size=15,vjust = -1), axis.title.y = element_text(size=15,vjust=4), axis.text = element_text(size=13))+ xlab(param$MAR.drawplot[j]) print(g1) } } } cat(paste(paste("The missing rate for MAR is", length(missing)/n, sep = " "),"\n",sep="")) } if(type=="MNAR"){ # requirement: param$data must be a data frame # param$MAR.type in c("increasing","tail","middle") # param$coeff # param$MAR.drawplot = T/F if(length(param)!=4){ stop("We require 4 parameters for MAR.") } if(!all(names(param) == c("data","MNAR.type","coeff","MNAR.drawplot"))){ stop("The input parameter must be [data],[MNAR.type],[coeff],[MNAR.drawplot]") } if(!is.data.frame(param$data) ){ stop("The data in param$data must be a data frame.") } if(nrow(param$data)!=n){ stop("The length of given dataset for MAR is wrong.\n") } if(!param$MNAR.type %in% c("increasing","tail","middle")){ stop("The MAR.type must be increasing/tail/middle.") } coeff <- param$coeff dat <- param$data %>% mutate(intercept=1) if(param$MNAR.type=="increasing"){ if(ncol(param$data) != (length(coeff)-1)){ stop("The number of coefficient for MNAR.type 'increasing' is wrong!") } if(ncol(dat)!=length(coeff)){ stop("The dat is not correct.") } ## missing probability ## for one variable logistic(coeff[1]*y_1+coeff[2]) p <- logistic(as.matrix(dat) %*% coeff) missing=(2:n)[rbinom((n-1),size=1,prob=(1-p[-1]))==0] if(printFlag){cat("The missing indicators for MNAR 'increasing' are generated!\n")} } if(param$MNAR.type=="tail"){ if(ncol(param$data) != (length(coeff)-2)){ stop("The number of coefficients for MNAR.type 'tail' is wrong.") } if(ncol(dat)!=length(coeff)-1){ stop("The dat is not correct.") } ## missing probability ## for one variable logistic(coeff[1]+abs(coeff[2]*y_1+coeff[3])) p <- logistic(coeff[1]+abs(as.matrix(dat) %*% coeff[-1])) missing=(2:n)[rbinom((n-1),size=1,prob=(1-p[-1]))==0] if(printFlag){cat("The missing indicators for MNAR 'tail' are generated!\n")} } if(param$MNAR.type=="middle"){ if(ncol(param$data) != (length(coeff)-2)){ stop("The number of coefficients for MNAR.type 'middle' is wrong.") } if(ncol(dat)!=length(coeff)-1){ stop("The dat is not correct.") } ## missing probability ## for one variable logistic(coeff[1]+abs(coeff[2]*y_1+coeff[3])) p <- logistic(coeff[1]-abs(as.matrix(dat) %*% coeff[-1])) missing=(2:n)[rbinom((n-1),size=1,prob=(1-p[-1]))==0] if(printFlag){cat("The missing indicators for MNAR 'middle' are generated!\n")} } if(printFlag){ if(param$MNAR.drawplot[1]) { # hist(p) for(m in 2:length(param$MNAR.drawplot)){ data_plot=data.frame(x=dat[-1,param$MNAR.drawplot[m]],p=p[-1]) g1 = ggplot(data_plot, aes(x=x, y=p))+ geom_line()+ theme_bw()+ ylab("Missing rates")+ theme(plot.margin = unit(c(1,1,1,1), "cm"), axis.title.x = element_text(size=15,vjust = -1), axis.title.y = element_text(size=15,vjust=4), axis.text = element_text(size=13))+ xlab(param$MNAR.drawplot[m]) print(g1) } } } cat(paste(paste("The missing rate for MNAR is", length(missing)/n, sep = " "),"\n",sep="")) } return(list(missing_index=missing,missing_rates=p)) }
ssmimputedemo::data_stationary data = data_stationary data$y_1 = c(NA,data$y[1:999]) data$x_1 = c(NA,data$x[1:999]) data$c_1 = c(NA,data$c[1:999]) index_MCAR = generate.missing_index(type = "MCAR", n=length(data$y), param = list(p=0.2))$missing_index index_MAR = generate.missing_index(type = "MAR", n=length(data$y), param =list(data = data.frame(y_1=data$y_1), MAR.type = "increasing", coeff = c(0.2,-8), MAR.drawplot = c(TRUE, "y_1")) )$missing_index index_MNAR = generate.missing_index(type = "MNAR", n=length(data$y), param = list(data = data.frame(y=data$y,x=data$x), MNAR.type = "increasing", coeff = c(0.5,-1,0), MNAR.drawplot = c(TRUE, "y")) )$missing_index
data$y[index_MAR] = NA data$y_1 = c(NA,data$y[1:999]) data$x_1 = c(NA,data$x[1:999]) data$c_1 = c(NA,data$c[1:999]) data = data[c(2:1000),] data_space_SSMimpute = data kable(head(data_space_SSMimpute)) imputeTS::ggplot_na_distribution(data_stationary$y, color_missing = "white",color_missing_border = "white", alpha_missing = 0.1) imputeTS::ggplot_na_distribution(data_space_SSMimpute$y, color_missing = "white",color_missing_border = "white", alpha_missing = 0.1)
printFlag=F formula="y~y_1+x+x_1+c" formula_var=unlist(strsplit(unlist(strsplit(formula,"~"))[2],"+",fixed=T)) ss_param=list(inits=c(log(0.25),log(1)),m0=c(40,0.5,-1,-0.5,-1),C0=diag(rep(10^3),5), AR1_coeffi=NULL,rw_coeffi="intercept", v_cp_param=NULL, w_cp_param=NULL,max_iteration=100) head(data_space_SSMimpute) result_statespace_SSMimpute1=run.SSMimpute_unanimous_cpts(data_ss_ori=data_space_SSMimpute,formula_var,ss_param_temp=ss_param, initial_imputation_option="StructTS", estimate_convergence_cri=0.01, lik_convergence_cri=0.01, stepsize_for_newpart=1/3, max_iteration=100, cpt_learning_param=list(cpt_method="mean",burnin=1/10,mergeband=20,convergence_cri=10), cpt_initial_guess_option="ignore", dlm_option="smooth",m=5,seed=1,printFlag=F)
#kable(result_statespace_SSMimpute1$result_convergence) #kable(result_statespace_SSMimpute1$result_convergence_mp) kable(result_statespace_SSMimpute1$result_convergence_mp_addV) result_statespace_SSMimpute1$estimated_cpts data_na = result_statespace_SSMimpute1$data_temp length(data_na$y_1) data_temp = result_statespace_SSMimpute1$data_temp missing_part=which(is.na(data_temp$y))[which(is.na(data_temp$y))<nrow(data_temp)] data_temp$y_1[missing_part+1]=result_statespace_SSMimpute1$y_final imputeTS::ggplot_na_distribution(data_stationary$y[1:999], color_missing = "pink",color_missing_border = "pink", alpha_missing = 0.9) imputeTS::ggplot_na_imputations(x_with_na = data_space_SSMimpute$y_1, x_with_imputations = data_na$y_1,x_with_truth = data_stationary$y[1:999])
R Package Documentation
Browse R Packages
We want your feedback!
Note that we can't provide technical support on individual packages. You should contact the package authors for that.
Embedding an R snippet on your website
Add the following code to your website.
For more information on customizing the embed code, read Embedding Snippets.