ADN
colSums(is.na(data_ADN))
## Provinsi Tahun Y X1 X2 X3 X4 X5
## 0 0 0 0 0 0 0 0
pred_use <- c("X1","X2","X3","X4","X5")
df_ADN2 <- data_ADN
# Transformasi Y (count data)
df_ADN2$t_Y <- log(df_ADN2$Y + 1)
# Transformasi X
for (v in pred_use) {
min_val <- min(df_ADN2[[v]], na.rm = TRUE)
if (min_val <= 0) {
shift <- abs(min_val) + 1
df_ADN2[[paste0("t_", v)]] <- log(df_ADN2[[v]] + shift)
} else {
df_ADN2[[paste0("t_", v)]] <- log(df_ADN2[[v]])
}
}
df_ADN2 <- na.omit(df_ADN2)
df_ADN2 %>%
select(t_X1, t_X2, t_X3, t_X4, t_X5) %>%
cor() %>%
ggcorrplot(type = "lower", lab = TRUE)
## Warning: `aes_string()` was deprecated in ggplot2 3.0.0.
## ℹ Please use tidy evaluation idioms with `aes()`.
## ℹ See also `vignette("ggplot2-in-packages")` for more information.
## ℹ The deprecated feature was likely used in the ggcorrplot package.
## Please report the issue at <https://github.com/kassambara/ggcorrplot/issues>.
## This warning is displayed once every 8 hours.
## Call `lifecycle::last_lifecycle_warnings()` to see where this warning was
## generated.
data_ADN %>% select(c(-1,-2)) %>% cor() %>%
ggcorrplot(type = "lower",lab = TRUE)
lm(t_Y ~ t_X1 + t_X2 + t_X3 + t_X4 + t_X5,
data = df_ADN2) %>% vif()
## t_X1 t_X2 t_X3 t_X4 t_X5
## 1.095283 1.794282 1.851701 1.159497 1.024614
cem <- plm(
t_Y ~ t_X1 + t_X2 + t_X3 + t_X4 + t_X5,
data = df_ADN2,
index = c("Provinsi","Tahun"),
model = "pooling"
)
fem <- plm(
t_Y ~ t_X1 + t_X2 + t_X3 + t_X4 + t_X5,
data = df_ADN2,
index = c("Provinsi","Tahun"),
model = "within"
)
rem <- plm(
t_Y ~ t_X1 + t_X2 + t_X3 + t_X4 + t_X5,
data = df_ADN2,
index = c("Provinsi","Tahun"),
model = "random"
)
pooltest(cem, fem)
##
## F statistic
##
## data: t_Y ~ t_X1 + t_X2 + t_X3 + t_X4 + t_X5
## F = 3.0704, df1 = 9, df2 = 55, p-value = 0.004726
## alternative hypothesis: unstability
phtest(fem, rem)
##
## Hausman Test
##
## data: t_Y ~ t_X1 + t_X2 + t_X3 + t_X4 + t_X5
## chisq = 3.2088, df = 5, p-value = 0.6678
## alternative hypothesis: one model is inconsistent
rem.final <- plm(
t_Y ~ t_X1 + t_X2 + t_X3 + t_X4 + t_X5,
data = df_ADN2,
index = c("Provinsi", "Tahun"),
model = "random"
)
summary(rem.final)
## Oneway (individual) effect Random Effect Model
## (Swamy-Arora's transformation)
##
## Call:
## plm(formula = t_Y ~ t_X1 + t_X2 + t_X3 + t_X4 + t_X5, data = df_ADN2,
## model = "random", index = c("Provinsi", "Tahun"))
##
## Balanced Panel: n = 10, T = 7, N = 70
##
## Effects:
## var std.dev share
## idiosyncratic 0.2806 0.5297 0.649
## individual 0.1515 0.3893 0.351
## theta: 0.5426
##
## Residuals:
## Min. 1st Qu. Median 3rd Qu. Max.
## -1.15924 -0.32467 0.07931 0.35244 1.16264
##
## Coefficients:
## Estimate Std. Error z-value Pr(>|z|)
## (Intercept) -5.757837 2.969785 -1.9388 0.0525250 .
## t_X1 -0.014967 0.053812 -0.2781 0.7809021
## t_X2 0.407158 0.112771 3.6105 0.0003056 ***
## t_X3 0.379748 0.375461 1.0114 0.3118157
## t_X4 0.431680 0.088390 4.8838 1.041e-06 ***
## t_X5 0.669994 0.326114 2.0545 0.0399294 *
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## Total Sum of Squares: 32.174
## Residual Sum of Squares: 17.548
## R-Squared: 0.4546
## Adj. R-Squared: 0.41199
## Chisq: 53.3456 on 5 DF, p-value: 2.8562e-10
KHP
colSums(is.na(data_KHP))
## Provinsi Tahun Y X1 X2 X3 X4 X5
## 0 0 0 0 0 0 0 0
pred_use <- c("X1","X2","X3","X4","X5")
data_KHP2 <- data_KHP
# Transformasi Y (count data)
data_KHP2$t_Y <- log(data_KHP2$Y + 1)
# Transformasi X
for (v in pred_use) {
min_val <- min(data_KHP2[[v]], na.rm = TRUE)
if (min_val <= 0) {
shift <- abs(min_val) + 1
data_KHP2[[paste0("t_", v)]] <- log(data_KHP2[[v]] + shift)
} else {
data_KHP2[[paste0("t_", v)]] <- log(data_KHP2[[v]])
}
}
data_KHP2 <- na.omit(data_KHP2)
data_KHP2 %>%
select(t_X1, t_X2, t_X3, t_X4, t_X5) %>%
cor() %>%
ggcorrplot(type = "lower", lab = TRUE)
data_KHP %>% select(c(-1,-2)) %>% cor() %>%
ggcorrplot(type = "lower",lab = TRUE)
lm(t_Y ~ t_X1 + t_X2 + t_X3 + t_X4 + t_X5,
data = data_KHP2) %>% vif()
## t_X1 t_X2 t_X3 t_X4 t_X5
## 1.332399 1.834779 2.088518 1.159603 1.178061
cem <- plm(
t_Y ~ t_X1 + t_X2 + t_X3 + t_X4 + t_X5,
data = data_KHP2,
index = c("Provinsi","Tahun"),
model = "pooling"
)
fem <- plm(
t_Y ~ t_X1 + t_X2 + t_X3 + t_X4 + t_X5,
data = data_KHP2,
index = c("Provinsi","Tahun"),
model = "within"
)
rem <- plm(
t_Y ~ t_X1 + t_X2 + t_X3 + t_X4 + t_X5,
data = data_KHP2,
index = c("Provinsi","Tahun"),
model = "random"
)
pooltest(cem, fem)
##
## F statistic
##
## data: t_Y ~ t_X1 + t_X2 + t_X3 + t_X4 + t_X5
## F = 3.0866, df1 = 9, df2 = 55, p-value = 0.004551
## alternative hypothesis: unstability
phtest(fem, rem)
##
## Hausman Test
##
## data: t_Y ~ t_X1 + t_X2 + t_X3 + t_X4 + t_X5
## chisq = 5.5712, df = 5, p-value = 0.3502
## alternative hypothesis: one model is inconsistent
rem.final <- plm(
t_Y ~ t_X1 + t_X2 + t_X3 + t_X4 + t_X5,
data = data_KHP2,
index = c("Provinsi", "Tahun"),
model = "random"
)
summary(rem.final)
## Oneway (individual) effect Random Effect Model
## (Swamy-Arora's transformation)
##
## Call:
## plm(formula = t_Y ~ t_X1 + t_X2 + t_X3 + t_X4 + t_X5, data = data_KHP2,
## model = "random", index = c("Provinsi", "Tahun"))
##
## Balanced Panel: n = 10, T = 7, N = 70
##
## Effects:
## var std.dev share
## idiosyncratic 0.2714 0.5209 0.675
## individual 0.1307 0.3615 0.325
## theta: 0.5217
##
## Residuals:
## Min. 1st Qu. Median 3rd Qu. Max.
## -1.096134 -0.288488 0.066527 0.347317 1.119853
##
## Coefficients:
## Estimate Std. Error z-value Pr(>|z|)
## (Intercept) -7.013334 3.109736 -2.2553 0.024116 *
## t_X1 0.099003 0.107370 0.9221 0.356494
## t_X2 0.420673 0.109260 3.8502 0.000118 ***
## t_X3 0.487862 0.379935 1.2841 0.199119
## t_X4 0.446294 0.088095 5.0661 4.061e-07 ***
## t_X5 0.600017 0.330150 1.8174 0.069154 .
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## Total Sum of Squares: 33.127
## Residual Sum of Squares: 17.496
## R-Squared: 0.47186
## Adj. R-Squared: 0.4306
## Chisq: 57.1799 on 5 DF, p-value: 4.6432e-11
library(plm)
pbgtest(rem.final)
##
## Breusch-Godfrey/Wooldridge test for serial correlation in panel models
##
## data: t_Y ~ t_X1 + t_X2 + t_X3 + t_X4 + t_X5
## chisq = 5.639, df = 7, p-value = 0.5825
## alternative hypothesis: serial correlation in idiosyncratic errors
library(lmtest)
bptest(rem.final)
##
## studentized Breusch-Pagan test
##
## data: rem.final
## BP = 3.4432, df = 5, p-value = 0.632
pcdtest(rem.final, test = "cd")
##
## Pesaran CD test for cross-sectional dependence in panels
##
## data: t_Y ~ t_X1 + t_X2 + t_X3 + t_X4 + t_X5
## z = -0.47588, p-value = 0.6342
## alternative hypothesis: cross-sectional dependence