-
Notifications
You must be signed in to change notification settings - Fork 0
/
R Code Sample_Regression.R
350 lines (276 loc) · 12 KB
/
R Code Sample_Regression.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
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
rm(list = ls())
setwd("C:/Users/huhu/Desktop/重构code/code/R code/Standard/R data A1")
dat <- read.csv("C:/Users/huhu/Desktop/重构code/code/R code/Standard/R data A1/R_vcm.csv")
library(np)
library(gtools)
library(fANCOVA)
library(foreign)
#source("npscoefGrad.R")
### Generating Data
yd <- dat[,1] #isei
yp <- dat[,8] #parent's log income (lnadinc)
ypx <- dat[,8] #parent's log incodatme (lnadinc)
age <- dat[,2] #child's age
gender <- dat[,3] #gender
urban <- dat[,4] #urban
minzu <- dat[,5] #minzu
mid <- dat[,6] #mid
east <- dat[,7] #east
ypz <- dat[,8] #parent's income (non logged)
ypze <- exp(dat[,8]) #parent's income (non logged)
basic<-data.frame(yd = yd, ypx=ypx, age=age, gender=gender, urban=urban, minzu=minzu, mid=mid, east=east, ypz=ypz, ypze=ypze)
#######################################################################
# Dropping missing obs
datall <- data.frame(basic=basic)
datall <-na.exclude(datall)
n <- dim(datall)[1]
#########################################################################################################################################################
yd <- datall[,1] #daughter's log income (lnayinc)
yd <- as.vector(yd)
ypx <- datall[,2] #parent's log income (lnadinc)
age <- datall[,3] #age
gender <- datall[,4] #gender
urban <- datall[,5] #urban
minzu <- datall[,6] #minzu
mid <- datall[,7] #mid
east <- datall[,8] #east
ypz <- datall[,9] #parent's income (not in logs)
ypze <- datall[,10] #parent's income (not in logs)
basic<-data.frame(yd = yd, ypx=ypx, age=age, gender=gender, urban=urban, minzu=minzu, mid=mid, east=east, ypz=ypz, ypze=ypze)
###################
########################
# value <-runif(1000)
# temp <- data.frame(value=value, quartile=rep(NA, 1000))
#############################################################################################################
x_basic <- data.frame(ypx=ypx, age=age, gender=gender, urban=urban, minzu=minzu, mid=mid, east=east)
########################################################################################################################
############### Estimation of IGM #########################################
################################################################################
# put together the data for estimation
###########################################
#ypz <- exp(ypz)/1000
###########################################
x_dat <- x_basic
z_dat <- ypz
yxz_dat <- cbind(yd,x_dat,ypz)
n <- nrow(yxz_dat)
###########################################################################
################ Linear model
lm_model <- lm(yd~ypx+age+gender+urban+minzu+mid+east,data=yxz_dat)
summary(lm_model)
lm_model_coef <- lm_model$coefficients
IGM_c <- lm_model_coef[2]
save(lm_model,file="lm_model.Rda")
#########################################################################
## Varying Coefficient Model
bw_cv <- npscoefbw(yd~ypx+age+gender+urban+minzu+mid+east|ypz, data=yxz_dat, nmulti=min(5,ncol(x_dat)+1))
bw <- npscoefbw(yd~ypx+age+gender+urban+minzu+mid+east|ypz, data=yxz_dat, nmulti=min(5,ncol(x_dat)+1),bandwidth.compute=FALSE,bws=2.5*bw_cv$bw)
vcm_model <- npscoef(bw,betas=TRUE)
gammas <- data.frame(vcm_model$beta)
trim_up <-0.01
trim_low <-0.01
temp <-0
temp <-cbind(z_dat,rep(lm_model_coef[2],n),gammas[,2])
temp <- temp[order(temp[,1]),]
matplot(temp[floor(1+n*trim_low):floor(n-n*trim_up),1],temp[floor(1+n*trim_low):floor(n-n*trim_up),2:3],type="llll",lty=c(2,1,3,3),pch=1, col = c(3,1), xlab="Parents' LogIncome", ylab="Coef.")
save(gammas,file="gammas.Rda")
# Computing the average VCM
avcm_coef <- colMeans(coef(vcm_model))
round(avcm_coef,3)
# Computing the quintiles of VCM sorted by parents' income (ypz)
tempor <-cbind(ypz,gammas)
tempor <- tempor[order(tempor[,1]),]
mgamma_q <-matrix(NA,5,dim(gammas)[2])
d_q1[,j]
for(j in 1:dim(gammas)[2]) {
temp <- data.frame(zval = tempor[,1], vcoef.value=tempor[,j+1], quintile=rep(NA, n))
brks <- with(temp, quantile(zval, probs = c(0, 0.2, 0.4, 0.6, 0.8, 1)))
temp <- within(temp, quintile <- cut(zval, breaks = brks, labels = 1:5,include.lowest = TRUE))
mgamma_q1<-mean(temp[temp[,3]==1,2])
mgamma_q2<-mean(temp[temp[,3]==2,2])
mgamma_q3<-mean(temp[temp[,3]==3,2])
mgamma_q4<-mean(temp[temp[,3]==4,2])
mgamma_q5<-mean(temp[temp[,3]==5,2])
mgamma_q[,j]<-c(mgamma_q1,mgamma_q2,mgamma_q3,mgamma_q4,mgamma_q5)
}
save(mgamma_q,file="mgamma_q.Rda")
d1_2 <- mgamma_q[1,] - mgamma_q[2,]
d1_3 <- mgamma_q[1,] - mgamma_q[3,]
d1_4 <- mgamma_q[1,] - mgamma_q[4,]
d1_5 <- mgamma_q[1,] - mgamma_q[5,]
d2_3 <- mgamma_q[2,] - mgamma_q[3,]
d2_4 <- mgamma_q[2,] - mgamma_q[4,]
d2_5 <- mgamma_q[2,] - mgamma_q[5,]
d3_4 <- mgamma_q[3,] - mgamma_q[4,]
d3_5 <- mgamma_q[3,] - mgamma_q[5,]
d4_5 <- mgamma_q[4,] - mgamma_q[5,]
################################################################################################################################# Bootstrap Inference #############################################
##########################################################################################################
num_boot <- 500 # set this to 500 or 1000
########## keep the original data #############
yxz_dat_orig <- yxz_dat
save(yxz_dat,file="yxz_dat.Rda")
x_dat_orig <- x_dat
z_dat_orig <- z_dat
############################################################################
############# Bootstrap Test the null of linear model vs. VCM (wild) ####################
set.seed(123)
fit0 <- fitted(lm_model)
res0 <- residuals(lm_model)
rss0 <- sum(res0^2)
res1 <-residuals(vcm_model) # residuals under H1: VCM
rss1 <-sum(res1^2)
GLRstat_boot <- matrix(NA,num_boot,1)
#GLRstat <- (n/2)*log(rss0/rss1)
GLRstat <- (n/2)*((rss0-rss1)/rss1)
res0_orig <- res0 - mean(res0)
res_mat_boot <- wild.boot(res0_orig, nboot=num_boot)
y_mat_boot <- matrix(rep(fit0,num_boot), ncol=num_boot) + res_mat_boot
for(b in 1:num_boot) {
#print(b)
# Generate boot sample: yd_star, xx_basic
yd <- y_mat_boot[,b]
yxz_dat <- cbind(yd,x_dat,ypz)
# Obtain rss0 under the null of the linear model
lm_model_boot <- lm(yd~ypx+age+gender+urban+minzu+mid+east,data=yxz_dat)
res0_boot <- residuals(lm_model_boot)
rss0_boot <- sum(res0_boot^2)
# Obtain rss1 under the alternative of the VCM
vcm_model_boot <- npscoef(bw,betas=TRUE)
res1_boot <- residuals(vcm_model_boot)
rss1_boot <- sum(res1_boot^2)
#GLRstat_boot[b,1] <- (n/2)*log(rss0_boot/rss1_boot)
GLRstat_boot[b,1] <- (n/2)*((rss0_boot-rss1_boot)/rss1_boot)
}
pval <- (sum(GLRstat_boot >= GLRstat)+1)/(num_boot+1)
print("P-value (wild) for testing the null of the linear against the alternative of VCM:")
print(pval)
saveRDS(pval,"pval1.Rda")
save(pval,file="teststat_pval.Rda")
######################################################################################
###############################################################################################
###### Wild bootstrap for std errors and CI of the VCM coefficients ############
set.seed(123)
res_vcm <- residuals(vcm_model) # VCM residuals
fit_vcm <- fitted(vcm_model)
res_vcm <- res_vcm - mean(res_vcm)
res_mat_boot <- wild.boot(res_vcm, nboot=num_boot)
y_mat_boot <- matrix(rep(fit_vcm,num_boot), ncol=num_boot) + res_mat_boot
# Generate objects (the number of objects should be the same as the number of coefficients + 1 for the avgcoef)
coef_mat_boot_1 <-matrix(NA,n,num_boot)
coef_mat_boot_2 <-matrix(NA,n,num_boot)
coef_mat_boot_3 <-matrix(NA,n,num_boot)
coef_mat_boot_4 <-matrix(NA,n,num_boot)
coef_mat_boot_5 <-matrix(NA,n,num_boot)
coef_mat_boot_6 <-matrix(NA,n,num_boot)
coef_mat_boot_7 <-matrix(NA,n,num_boot)
coef_mat_boot_8 <-matrix(NA,n,num_boot)
mgamma_q_boot_1 <-matrix(NA,5,num_boot)
mgamma_q_boot_2 <-matrix(NA,5,num_boot)
mgamma_q_boot_3 <-matrix(NA,5,num_boot)
mgamma_q_boot_4 <-matrix(NA,5,num_boot)
mgamma_q_boot_5 <-matrix(NA,5,num_boot)
mgamma_q_boot_6 <-matrix(NA,5,num_boot)
mgamma_q_boot_7 <-matrix(NA,5,num_boot)
mgamma_q_boot_8 <-matrix(NA,5,num_boot)
d1_2_boot <- matrix(NA,8,num_boot)
d1_2_boot <- matrix(NA,8,num_boot)
d1_3_boot <- matrix(NA,8,num_boot)
d1_4_boot <- matrix(NA,8,num_boot)
d1_5_boot <- matrix(NA,8,num_boot)
d2_3_boot <- matrix(NA,8,num_boot)
d2_4_boot <- matrix(NA,8,num_boot)
d2_5_boot <- matrix(NA,8,num_boot)
d3_4_boot <- matrix(NA,8,num_boot)
d3_5_boot <- matrix(NA,8,num_boot)
d4_5_boot <- matrix(NA,8,num_boot)
k <- ncol(gammas)
coef_mat_boot_avg <-matrix(NA,k,num_boot)
for(b in 1:num_boot) {
# print(b)
yd <- y_mat_boot[,b]
yxz_dat <- cbind(yd,x_dat,ypz)
vcm_model_boot <- npscoef(bws=bw,betas=TRUE)
gammas_boot <- data.frame(vcm_model_boot$beta)
coef_mat_boot_1[,b] <- gammas_boot[,1]
coef_mat_boot_2[,b] <- gammas_boot[,2]
coef_mat_boot_3[,b] <- gammas_boot[,3]
coef_mat_boot_4[,b] <- gammas_boot[,4]
coef_mat_boot_5[,b] <- gammas_boot[,5]
coef_mat_boot_6[,b] <- gammas_boot[,6]
coef_mat_boot_7[,b] <- gammas_boot[,7]
coef_mat_boot_8[,b] <- gammas_boot[,8]
coef_mat_boot_avg[,b] <- t(colMeans(coef(vcm_model_boot)))
# Computing the quintiles of VCM sorted by parents' income (ypz)
tempor <-cbind(ypz,gammas_boot)
tempor <- tempor[order(tempor[,1]),]
mgamma_q_boot <-matrix(NA,5,dim(gammas_boot)[2])
for(j in 1:dim(gammas)[2]) {
temp <- data.frame(zval = tempor[,1], vcoef.value=tempor[,j+1], quintile=rep(NA, n))
brks <- with(temp, quantile(zval, probs = c(0, 0.2, 0.4, 0.6, 0.8, 1)))
temp <- within(temp, quintile <- cut(zval, breaks = brks, labels = 1:5,include.lowest = TRUE))
mgamma_q1<-mean(temp[temp[,3]==1,2])
mgamma_q2<-mean(temp[temp[,3]==2,2])
mgamma_q3<-mean(temp[temp[,3]==3,2])
mgamma_q4<-mean(temp[temp[,3]==4,2])
mgamma_q5<-mean(temp[temp[,3]==5,2])
mgamma_q_boot[,j]<-c(mgamma_q1,mgamma_q2,mgamma_q3,mgamma_q4,mgamma_q5)
}
mgamma_q_boot_1[,b] <- mgamma_q_boot[,1]
mgamma_q_boot_2[,b] <- mgamma_q_boot[,2]
mgamma_q_boot_3[,b] <- mgamma_q_boot[,3]
mgamma_q_boot_4[,b] <- mgamma_q_boot[,4]
mgamma_q_boot_5[,b] <- mgamma_q_boot[,5]
mgamma_q_boot_6[,b] <- mgamma_q_boot[,6]
mgamma_q_boot_7[,b] <- mgamma_q_boot[,7]
mgamma_q_boot_8[,b] <- mgamma_q_boot[,8]
d1_2_boot[,b] <- mgamma_q_boot[1,] - mgamma_q_boot[2,]
d1_3_boot[,b] <- mgamma_q_boot[1,] - mgamma_q_boot[3,]
d1_4_boot[,b] <- mgamma_q_boot[1,] - mgamma_q_boot[4,]
d1_5_boot[,b] <- mgamma_q_boot[1,] - mgamma_q_boot[5,]
d2_3_boot[,b] <- mgamma_q_boot[2,] - mgamma_q_boot[3,]
d2_4_boot[,b] <- mgamma_q_boot[2,] - mgamma_q_boot[4,]
d2_5_boot[,b] <- mgamma_q_boot[2,] - mgamma_q_boot[5,]
d3_4_boot[,b] <- mgamma_q_boot[3,] - mgamma_q_boot[4,]
d3_5_boot[,b] <- mgamma_q_boot[3,] - mgamma_q_boot[5,]
d4_5_boot[,b] <- mgamma_q_boot[4,] - mgamma_q_boot[5,]
}
IGM <- data.frame(IGM = IGM)
write.csv(IGM,file = "IGM_Dau1.csv")
save(coef_mat_boot_1,file="coef_mat_boot_1.Rda")
save(coef_mat_boot_2,file="coef_mat_boot_2.Rda")
save(coef_mat_boot_3,file="coef_mat_boot_3.Rda")
save(coef_mat_boot_4,file="coef_mat_boot_4.Rda")
save(coef_mat_boot_5,file="coef_mat_boot_5.Rda")
save(coef_mat_boot_6,file="coef_mat_boot_6.Rda")
save(coef_mat_boot_7,file="coef_mat_boot_7.Rda")
save(coef_mat_boot_8,file="coef_mat_boot_8.Rda")
save(coef_mat_boot_avg,file="coef_mat_boot_avg.Rda")
save(mgamma_q_boot_1,file="mgamma_q_boot_1.Rda")
save(mgamma_q_boot_2,file="mgamma_q_boot_2.Rda")
save(mgamma_q_boot_3,file="mgamma_q_boot_3.Rda")
save(mgamma_q_boot_4,file="mgamma_q_boot_4.Rda")
save(mgamma_q_boot_5,file="mgamma_q_boot_5.Rda")
save(mgamma_q_boot_6,file="mgamma_q_boot_6.Rda")
save(mgamma_q_boot_7,file="mgamma_q_boot_7.Rda")
save(mgamma_q_boot_8,file="mgamma_q_boot_8.Rda")
save(d1_2_boot,file="d1_2_boot.Rda")
save(d1_3_boot,file="d1_3_boot.Rda")
save(d1_4_boot,file="d1_4_boot.Rda")
save(d1_5_boot,file="d1_5_boot.Rda")
save(d2_3_boot,file="d2_3_boot.Rda")
save(d2_4_boot,file="d2_4_boot.Rda")
save(d2_5_boot,file="d2_5_boot.Rda")
save(d3_4_boot,file="d3_4_boot.Rda")
save(d3_5_boot,file="d3_5_boot.Rda")
save(d4_5_boot,file="d4_5_boot.Rda")
save(d1_2,file="d1_2.Rda")
save(d1_3,file="d1_3.Rda")
save(d1_4,file="d1_4.Rda")
save(d1_5,file="d1_5.Rda")
save(d2_3,file="d2_3.Rda")
save(d2_4,file="d2_4.Rda")
save(d2_5,file="d2_5.Rda")
save(d3_4,file="d3_4.Rda")
save(d3_5,file="d3_5.Rda")
save(d4_5,file="d4_5.Rda")