Multinomial

library(haven)
datos <- haven::read_spss("GSS2006.sav", col_select = c("NATMASS","AGE", "SEX", "SEI", "REGION", "POLVIEWS"))

  1. Convierta la variable NATMASS a factor cuyo nivel base sea about right

  2. Recodifique la variable POLVIEWS para que sea un factor que vaya desde extremely liberal hasta extremely conservative

  3. ¿Qué inclinación política es más frecuente?, diseñe un gráfico de inclinación política y gasto en transporte masivo (NATMASS) y describa detalladamente.

Hay un gasto en transoporte modico

table1 <- table(datos$POLVIEWS)
barplot(table1, xlab='Inclinación política', ylab='Frecuencia', las=1)

Existe un mayor numero de personas que tienen una posición politica moderada, ni tan conservadora ni tan liberal

table2 <- table(datos$NATMASS)
barplot(table1, xlab='Inclinación política', ylab='Frecuencia', las=1)

Modelos

Modelo 1

library(nnet)
mod1 <- multinom(NATMASS~AGE+SEX+SEI+REGION, data=datos)
## # weights:  18 (10 variable)
## initial  value 2873.969747 
## iter  10 value 2472.960196
## final  value 2400.202006 
## converged
print(mod1)
## Call:
## multinom(formula = NATMASS ~ AGE + SEX + SEI + REGION, data = datos)
## 
## Coefficients:
##   (Intercept)          AGE           SEX         SEI       REGION
## 2   0.9823396 -0.009927595  0.2928898089 -0.01596910 -0.019428425
## 3  -0.7939748 -0.002324534 -0.0002714376 -0.01390107  0.009000034
## 
## Residual Deviance: 4800.404 
## AIC: 4820.404

Modelo 2

mod2 <- multinom(POLVIEWS~AGE+SEX+SEI+REGION, data=datos)
## # weights:  42 (30 variable)
## initial  value 7931.529768 
## iter  10 value 7350.519410
## iter  20 value 7258.544420
## iter  30 value 6822.795859
## iter  40 value 6788.621778
## iter  40 value 6788.621754
## iter  40 value 6788.621754
## final  value 6788.621754 
## converged
print(mod2)
## Call:
## multinom(formula = POLVIEWS ~ AGE + SEX + SEI + REGION, data = datos)
## 
## Coefficients:
##   (Intercept)         AGE         SEX           SEI       REGION
## 2   0.7281361 0.006944577  0.09559675  0.0076928231 -0.042503151
## 3   1.0782476 0.005886851 -0.15394581  0.0061286399 -0.013137867
## 4   2.5345922 0.012282620  0.02020009 -0.0074829566 -0.044325150
## 5   1.0753909 0.012602818 -0.13501355  0.0029489031 -0.008876476
## 6   1.1720314 0.021738579 -0.34640746  0.0005575264 -0.013533450
## 7  -0.6495382 0.029690338 -0.14726936 -0.0067877767  0.004849077
## 
## Residual Deviance: 13577.24 
## AIC: 13637.24

Residuales

summary(mod2)
## Call:
## multinom(formula = POLVIEWS ~ AGE + SEX + SEI + REGION, data = datos)
## 
## Coefficients:
##   (Intercept)         AGE         SEX           SEI       REGION
## 2   0.7281361 0.006944577  0.09559675  0.0076928231 -0.042503151
## 3   1.0782476 0.005886851 -0.15394581  0.0061286399 -0.013137867
## 4   2.5345922 0.012282620  0.02020009 -0.0074829566 -0.044325150
## 5   1.0753909 0.012602818 -0.13501355  0.0029489031 -0.008876476
## 6   1.1720314 0.021738579 -0.34640746  0.0005575264 -0.013533450
## 7  -0.6495382 0.029690338 -0.14726936 -0.0067877767  0.004849077
## 
## Std. Errors:
##   (Intercept)         AGE       SEX         SEI     REGION
## 2   0.5477045 0.006347857 0.2020383 0.005111932 0.04111714
## 3   0.5462869 0.006366788 0.2017307 0.005123405 0.04114729
## 4   0.5059197 0.005897294 0.1871577 0.004783067 0.03814638
## 5   0.5366978 0.006227436 0.1981144 0.005039448 0.04039261
## 6   0.5333506 0.006176661 0.1970262 0.005016426 0.04019432
## 7   0.6571688 0.007346574 0.2405974 0.006185675 0.04906914
## 
## Residual Deviance: 13577.24 
## AIC: 13637.24
plot(mod2$residuals)