-
Notifications
You must be signed in to change notification settings - Fork 1
/
Copy pathestimate.R
169 lines (133 loc) · 6.36 KB
/
estimate.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
setwd("~/binary_mrt/lag-effect-1")
source("estimators.R")
source("estimators_robust_adhocery.R") #didn't use the function in this file
source("data_generating.R")
data_generating_process <- dgm_binary_categorical_covariate
# library(tidyverse)
library(foreach)
library(doMC)
library(doRNG)
compute_result_beta <- function(beta_true, beta, beta_se, beta_se_adjusted, moderator_vars, control_vars, significance_level,
na.rm = FALSE) {
beta_true_array <- matrix(NA, nrow = nrow(beta),ncol = length(beta_true))
for (ind1 in 1:dim(beta_true_array)[1]) {
beta_true_array[ind1,] <- beta_true
}
p <- length(moderator_vars) + 1
q <- length(control_vars) + 1
bias <- mean(beta - beta_true_array,na.rm = na.rm)
sd <- sd(beta, na.rm = na.rm)
rmse <- sqrt(mean((beta - beta_true_array)^2))
critical_factor <- qnorm(1 - significance_level/2)
ci_left <- beta - critical_factor * beta_se
ci_right <- beta + critical_factor * beta_se
beta_se_mean = mean(beta_se)
coverage_prob <- mean((ci_left < beta_true_array) & (ci_right > beta_true_array))
critical_factor_adj <- qt(1 - significance_level/2, df = sample_size - 1 - q)
ci_left_adj <- beta - critical_factor_adj * beta_se_adjusted
ci_right_adj <- beta + critical_factor_adj * beta_se_adjusted
beta_se_adjusted_mean = mean(beta_se_adjusted)
coverage_prob_adj <- mean((ci_left_adj < beta_true_array) & (ci_right_adj > beta_true_array))
return(list(bias = bias,se=beta_se_mean, se_adjusted=beta_se_adjusted_mean,
sd = sd, rmse = rmse, coverage_prob = coverage_prob, coverage_prob_adjusted = coverage_prob_adj))
}
max_cores <- 4
registerDoMC(min(detectCores() - 1, max_cores))
# sample_sizes <- c(250, 625, 1000)
total_T <- 30
nsim <- 1000
control_vars <- "S"
moderator_vars <- c()
result_df_collected_1 <- data.frame()
result_df_collected_2 <- data.frame()
for (i_ss in 1:6) {
group_ls = group_all[[i_ss]]
sample_size <- group_ls[["n"]]
result <- foreach(isim = 1:nsim, .combine = "c") %dorng% {
if (isim %% 100 == 0) {
cat(paste("Starting iteration",isim,"\n"))
}
dta <- data_generating_process(sample_size, total_T,group_ls)
fit_wcls <- weighted_centered_least_square(
dta = dta,
group_ls=group_ls,
id_varname = "userid",
decision_time_varname = "day",
treatment_varname = "A",
outcome_varname = "Y_delta",
control_varname = control_vars,
moderator_varname = moderator_vars,
rand_prob_varname = "prob_A",
rand_prob_tilde_varname = NULL,
rand_prob_tilde = 0.2,
estimator_initial_value = NULL,
future_treatment = "ST"
)
fit_wcls_2 <- weighted_centered_least_square(
dta = dta,
group_ls=group_ls,
id_varname = "userid",
decision_time_varname = "day",
treatment_varname = "A",
outcome_varname = "Y_delta",
control_varname = control_vars,
moderator_varname = moderator_vars,
rand_prob_varname = "prob_A",
rand_prob_tilde_varname = NULL,
rand_prob_tilde = 0.2,
estimator_initial_value = NULL,
future_treatment = "OTD"
)
output <- list(fit_wcls = fit_wcls,fit_wcls_2 = fit_wcls_2)
}
ee_names <- "wcls"
alpha_names <- c("Intercept", control_vars)
beta_names <- c("Intercept", moderator_vars)
num_estimator <- length(ee_names)
result_1 = result[seq(1,2*nsim-1,by=2)]
result_2 = result[seq(2,2*nsim,by=2)]
alpha <- matrix(sapply(result_1, function(l) l$alpha_hat), byrow = TRUE,nrow =nsim )
alpha_se <- matrix(sapply(result_1, function(l) l$alpha_se),byrow = TRUE,nrow =nsim)
alpha_se_adjusted <- matrix(sapply(result_1, function(l) l$alpha_se_adjusted),byrow = TRUE, nrow = nsim)
colnames(alpha)= colnames(alpha_se)= colnames(alpha_se_adjusted) = alpha_names
beta <- matrix(sapply(result_1, function(l) l$beta_hat))
beta_se <- matrix(sapply(result_1, function(l) l$beta_se))
beta_se_adjusted <- matrix(sapply(result_1, function(l) l$beta_se_adjusted))
colnames(beta)= colnames(beta_se) = colnames(beta_se_adjusted)= beta_names
result <- compute_result_beta(beta_true_marginal, beta, beta_se, beta_se_adjusted, moderator_vars, control_vars, significance_level = 0.05)
result_df <- data.frame(ss = rep(sample_size, num_estimator),
est = ee_names,
bias = result$bias,
se=result$se,
se_adjusted = result$se_adjusted,
sd = result$sd,
rmse = result$rmse,
cp.unadj = result$coverage_prob,
cp.adj = result$coverage_prob_adjusted)
names(result_df) <- c("ss", "est", "bias","se.unadj","se.adj", "sd", "rmse", "cp.unadj", "cp.adj")
rownames(result_df) <- NULL
result_df_collected_1 <- rbind(result_df_collected_1, result_df)
alpha <- matrix(sapply(result_2, function(l) l$alpha_hat), byrow = TRUE,nrow =nsim )
alpha_se <- matrix(sapply(result_2, function(l) l$alpha_se),byrow = TRUE,nrow =nsim)
alpha_se_adjusted <- matrix(sapply(result_2, function(l) l$alpha_se_adjusted),byrow = TRUE, nrow = nsim)
colnames(alpha)= colnames(alpha_se)= colnames(alpha_se_adjusted) = alpha_names
beta <- matrix(sapply(result_2, function(l) l$beta_hat))
beta_se <- matrix(sapply(result_2, function(l) l$beta_se))
beta_se_adjusted <- matrix(sapply(result_2, function(l) l$beta_se_adjusted))
colnames(beta)= colnames(beta_se) = colnames(beta_se_adjusted)= beta_names
result <- compute_result_beta(beta_true_marginal, beta, beta_se, beta_se_adjusted, moderator_vars, control_vars, significance_level = 0.05)
result_df <- data.frame(ss = rep(sample_size, num_estimator),
est = ee_names,
bias = result$bias,
se=result$se,
se_adjusted = result$se_adjusted,
sd = result$sd,
rmse = result$rmse,
cp.unadj = result$coverage_prob,
cp.adj = result$coverage_prob_adjusted)
names(result_df) <- c("ss", "est", "bias","se.unadj","se.adj", "sd", "rmse", "cp.unadj", "cp.adj")
rownames(result_df) <- NULL
result_df_collected_2 <- rbind(result_df_collected_2, result_df)
}
save(result_df_collected_1,file = "cwcls-st.RData")
save(result_df_collected_2,file = "cwcls-otd.RData")