suppressMessages(library(tidyverse))
suppressMessages(library(fpp2))
suppressMessages(library(psych))
suppressMessages(library(BMA))
suppressMessages(library(GGally))
setwd("C:/Users/DellPC/Desktop")
df <- read.csv("a.csv") %>% select (-1)Phan tich thong ke mo ta
describe(df)## vars n mean sd median trimmed mad min max range skew kurtosis
## Ln_FDI 1 30 7.98 1.27 7.95 7.93 1.10 5.54 11.02 5.48 0.32 -0.08
## Ln_GDP 2 30 11.68 1.40 11.64 11.64 1.54 9.48 14.82 5.34 0.20 -0.67
## Ln_Tr 3 30 23.56 1.52 23.27 23.45 1.91 21.48 26.71 5.24 0.43 -0.89
## Ln_Pc 4 30 6.88 0.82 7.20 6.92 0.84 4.80 8.23 3.43 -0.55 -0.43
## Ln_Wgr 5 30 3.84 0.44 3.92 3.89 0.31 2.66 4.44 1.78 -1.05 0.37
## Opn 6 30 0.72 0.41 0.59 0.68 0.35 0.11 2.08 1.97 1.21 1.68
## se
## Ln_FDI 0.23
## Ln_GDP 0.26
## Ln_Tr 0.28
## Ln_Pc 0.15
## Ln_Wgr 0.08
## Opn 0.08Correlation giua cac bien
ggpairs(df)
Hoi quy tuyen tinh
model <- lm(Ln_FDI ~ Ln_GDP + Ln_Pc + Ln_Tr+ Ln_Wgr + Opn, data = df)
summary(model)##
## Call:
## lm(formula = Ln_FDI ~ Ln_GDP + Ln_Pc + Ln_Tr + Ln_Wgr + Opn,
## data = df)
##
## Residuals:
## Min 1Q Median 3Q Max
## -1.4933 -0.3918 0.1294 0.4723 1.1476
##
## Coefficients:
## Estimate Std. Error t value Pr(>|t|)
## (Intercept) -2.98254 2.53819 -1.175 0.25149
## Ln_GDP 0.77235 0.22623 3.414 0.00228 **
## Ln_Pc 0.31268 0.24699 1.266 0.21768
## Ln_Tr 0.03367 0.18208 0.185 0.85483
## Ln_Wgr -0.46473 0.44079 -1.054 0.30223
## Opn 1.07810 0.42439 2.540 0.01796 *
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## Residual standard error: 0.7095 on 24 degrees of freedom
## Multiple R-squared: 0.7427, Adjusted R-squared: 0.6891
## F-statistic: 13.86 on 5 and 24 DF, p-value: 2.031e-06Kiem tra gia thuyet cua mo hinh
checkresiduals(model)
##
## Breusch-Godfrey test for serial correlation of order up to 9
##
## data: Residuals
## LM test = 4.5722, df = 9, p-value = 0.8699Sau khi chay mo hinh thi mo hinh chi co 2 bien thoa man duoc dieu kien cua hoi quy tuyen tinh
Chon mo hinh dua vao chi so BIC (Bayensian Information Criteria)
xvars <- df[,-1]
y= df[,1]
md <- bicreg(xvars, y, strict = FALSE, OR=20)
summary(md)##
## Call:
## bicreg(x = xvars, y = y, strict = FALSE, OR = 20)
##
##
## 5 models were selected
## Best 5 models (cumulative posterior probability = 1 ):
##
## p!=0 EV SD model 1 model 2 model 3 model 4
## Intercept 100.0 -2.98006 1.57458 -2.95565 -3.50264 -2.73312 -2.81323
## Ln_GDP 100.0 0.85452 0.12044 0.85883 0.83642 0.85951 0.87267
## Ln_Tr 10.7 -0.00143 0.05805 . . . -0.01334
## Ln_Pc 19.8 0.03378 0.11323 . 0.12347 . .
## Ln_Wgr 16.1 -0.02975 0.17014 . . -0.06020 .
## Opn 100.0 1.23057 0.35552 1.24358 1.18806 1.24495 1.25762
##
## nVar 2 3 3 3
## r2 0.725 0.731 0.726 0.725
## BIC -31.93695 -29.16654 -28.58162 -28.54230
## post prob 0.585 0.146 0.109 0.107
## model 5
## Intercept -2.64363
## Ln_GDP 0.80870
## Ln_Tr .
## Ln_Pc 0.30395
## Ln_Wgr -0.44869
## Opn 1.11712
##
## nVar 4
## r2 0.742
## BIC -27.08097
## post prob 0.052Chon model 5 la vi BIC lon nhat
model <- lm(Ln_FDI ~ Ln_GDP + Ln_Pc + Ln_Wgr + Opn, data = df)
summary(model)##
## Call:
## lm(formula = Ln_FDI ~ Ln_GDP + Ln_Pc + Ln_Wgr + Opn, data = df)
##
## Residuals:
## Min 1Q Median 3Q Max
## -1.4667 -0.4321 0.1216 0.4884 1.1699
##
## Coefficients:
## Estimate Std. Error t value Pr(>|t|)
## (Intercept) -2.6436 1.7219 -1.535 0.13728
## Ln_GDP 0.8087 0.1099 7.360 1.04e-07 ***
## Ln_Pc 0.3040 0.2377 1.279 0.21276
## Ln_Wgr -0.4487 0.4237 -1.059 0.29977
## Opn 1.1171 0.3611 3.094 0.00481 **
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## Residual standard error: 0.6956 on 25 degrees of freedom
## Multiple R-squared: 0.7424, Adjusted R-squared: 0.7011
## F-statistic: 18.01 on 4 and 25 DF, p-value: 4.462e-07