Regresiones

Regresión Gasto Público, Tasa Crecimiento

Importar datos 1

url1<-"https://raw.githubusercontent.com/armandovl/datasets_uno/main/GPA_Y_CRECIMIENTO.csv"
datos1<-read.csv(url(url1))
head(datos1)

Modelado 1

modelo1<-lm(Tasa_Crecimiento~Gasto_Publico, datos1)

summary(modelo1)

Call:
lm(formula = Tasa_Crecimiento ~ Gasto_Publico, data = datos1)

Residuals:
    Min      1Q  Median      3Q     Max 
-1.2914 -0.6642 -0.0423  0.2239  2.0047 

Coefficients:
              Estimate Std. Error t value Pr(>|t|)    
(Intercept)    0.73651    0.21572   3.414  0.00113 ** 
Gasto_Publico  0.31240    0.02834  11.023 3.03e-16 ***
---
Signif. codes:  0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1

Residual standard error: 0.9093 on 62 degrees of freedom
Multiple R-squared:  0.6621,    Adjusted R-squared:  0.6567 
F-statistic: 121.5 on 1 and 62 DF,  p-value: 3.026e-16

Supuesto linealidad 1 (la línea roja debe acercarse a la horizontal)

plot(modelo1,1)

plot(Tasa_Crecimiento~Gasto_Publico,datos1,
     pch=18,
     col="purple")

Supuesto media del error = 0 (también sirve creo para la linealidad)

residuos<-modelo1$residuals

mean(residuos)

[1] -2.072351e-17

supuesto normalidad de los errores para este caso no hay normalidad

shapiro.test(residuos)

    Shapiro-Wilk normality test

data:  residuos
W = 0.92014, p-value = 0.0005036

library(nortest)
ad.test (residuos)

    Anderson-Darling normality test

data:  residuos
A = 1.6519, p-value = 0.0002744

hist(residuos , breaks = 30)

supuestos homogeneidad de las varianzas

si son homogeneas

#En este caso se rechaza la h0, por lo tanto hay heteregeneidad de varianza,no se cumple el supuesto
library(lmtest)
bptest(modelo1)

    studentized Breusch-Pagan test

data:  modelo1
BP = 0.030721, df = 1, p-value = 0.8609

predichos=fitted(modelo1)
plot(residuos~predichos)

supuesto de error con X

plot (datos1$Gasto_Publico,residuos)

Regresión Cobb Douglas

url2<-"https://raw.githubusercontent.com/armandovl/datasets_uno/main/cobb_douglas.csv"
datos2<-read.csv(url(url2))
head(datos2,8)

NA

regresion1<-lm(produccion~capital+superficie+trabajadores,datos2)
summary(regresion1)

Call:
lm(formula = produccion ~ capital + superficie + trabajadores, 
    data = datos2)

Residuals:
     Min       1Q   Median       3Q      Max 
-1779.87 -1104.16   -64.34   961.48  1916.98 

Coefficients:
               Estimate Std. Error t value Pr(>|t|)   
(Intercept)  -7787.0699  6576.6682  -1.184  0.26135   
capital      10324.7667  2846.5727   3.627  0.00398 **
superficie       1.2729     0.4615   2.758  0.01862 * 
trabajadores     1.9611     0.8163   2.402  0.03509 * 
---
Signif. codes:  0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1

Residual standard error: 1362 on 11 degrees of freedom
Multiple R-squared:  0.9574,    Adjusted R-squared:  0.9458 
F-statistic: 82.39 on 3 and 11 DF,  p-value: 8.033e-08

regresion2<-lm(log(produccion)~log(capital)+log(superficie)+log(trabajadores),datos2)
summary(regresion2)

Call:
lm(formula = log(produccion) ~ log(capital) + log(superficie) + 
    log(trabajadores), data = datos2)

Residuals:
      Min        1Q    Median        3Q       Max 
-0.046143 -0.027828  0.001433  0.022655  0.043295 

Coefficients:
                  Estimate Std. Error t value Pr(>|t|)   
(Intercept)         2.9408     2.3406   1.256  0.23500   
log(capital)        0.3528     0.1028   3.431  0.00561 **
log(superficie)     0.5187     0.1944   2.668  0.02187 * 
log(trabajadores)   0.2870     0.1607   1.786  0.10162   
---
Signif. codes:  0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1

Residual standard error: 0.03512 on 11 degrees of freedom
Multiple R-squared:  0.9548,    Adjusted R-squared:  0.9425 
F-statistic: 77.51 on 3 and 11 DF,  p-value: 1.106e-07

dwtest(regresion2) #no hay autocorrelación

    Durbin-Watson test

data:  regresion2
DW = 1.9354, p-value = 0.2951
alternative hypothesis: true autocorrelation is greater than 0

Regresión Regiones

url3<-"https://raw.githubusercontent.com/armandovl/datasets_uno/main/regiones.csv"
datos3<-read.csv(url(url3))
head(datos3,8)

datos3$Region <- as.factor(datos3$Region) #convertir a factor

regresion3<-lm(Gasto~Region+Salario,datos3)

Warning in for (i in seq_len(n)) { :
  cerrando la conenexion 3 (https://raw.githubusercontent.com/armandovl/datasets_uno/main/regiones.csv) que no esta siendo utilizada 

summary(regresion3)

Call:
lm(formula = Gasto ~ Region + Salario, data = datos3)

Residuals:
     Min       1Q   Median       3Q      Max 
-1332.78  -318.78    -8.37   323.14  1295.64 

Coefficients:
              Estimate Std. Error t value Pr(>|t|)    
(Intercept) -1.262e+03  5.142e+02  -2.455   0.0179 *  
RegionOeste -3.481e+02  2.062e+02  -1.688   0.0980 .  
RegionSur   -3.028e+02  1.904e+02  -1.591   0.1184    
Salario      2.114e-01  2.042e-02  10.354 1.03e-13 ***
---
Signif. codes:  0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1

Residual standard error: 575.5 on 47 degrees of freedom
Multiple R-squared:  0.7201,    Adjusted R-squared:  0.7023 
F-statistic: 40.31 on 3 and 47 DF,  p-value: 4.791e-13

ggplot(datos3, aes(x=Salario, 
                   y=Gasto, 
                   color=Region, 
                   shape=Region, 
                   )
       )+ facet_grid(Region~.)+
  
  geom_point(size=2) + 
  
  geom_smooth(method=lm, se=FALSE, fullrange=TRUE)

Regresión Experiencia y Sexo

url4<-"https://raw.githubusercontent.com/armandovl/datasets_uno/main/ingreso.csv"
datos4<-read.csv(url(url4))
head(datos4,8)

regresion4<-lm(ingreso~genero+exp, datos4) #modelo regresión
summary(regresion4) #estadísticos del modelo

Call:
lm(formula = ingreso ~ genero + exp, data = datos4)

Residuals:
    Min      1Q  Median      3Q     Max 
-3629.6  -890.4  -273.9  1163.2  4166.6 

Coefficients:
            Estimate Std. Error t value Pr(>|t|)    
(Intercept) 11871.73     836.95  14.185 7.49e-11 ***
generomujer -4580.74     928.26  -4.935 0.000126 ***
exp           329.24      53.33   6.173 1.02e-05 ***
---
Signif. codes:  0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1

Residual standard error: 2011 on 17 degrees of freedom
Multiple R-squared:  0.7481,    Adjusted R-squared:  0.7185 
F-statistic: 25.24 on 2 and 17 DF,  p-value: 8.135e-06

ggplot(datos4, aes(x=exp, y=ingreso, color=genero, shape=genero)) +
  geom_point(size=2) + 
  geom_smooth(method=lm, se=FALSE, fullrange=TRUE)

Warning: cerrando la conenexion 3 (https://raw.githubusercontent.com/armandovl/datasets_uno/main/ingresos.csv) que no esta siendo utilizada 

regresion45<-lm(log(ingreso)~genero+log(exp), datos4) #modelo regresión
summary(regresion45) #estadísticos del modelo

Call:
lm(formula = log(ingreso) ~ genero + log(exp), data = datos4)

Residuals:
     Min       1Q   Median       3Q      Max 
-0.35093 -0.07041  0.00149  0.09529  0.29485 

Coefficients:
            Estimate Std. Error t value Pr(>|t|)    
(Intercept)  9.08757    0.11287  80.516  < 2e-16 ***
generomujer -0.38709    0.07956  -4.866 0.000145 ***
log(exp)     0.26217    0.05078   5.162 7.81e-05 ***
---
Signif. codes:  0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1

Residual standard error: 0.1658 on 17 degrees of freedom
Multiple R-squared:  0.685, Adjusted R-squared:  0.6479 
F-statistic: 18.48 on 2 and 17 DF,  p-value: 5.441e-05

Cobb_douglas 2

datos5<-read.csv(url("https://raw.githubusercontent.com/armandovl/datasets_uno/main/cobb2.csv"))
head(datos5,7) #estructura de los datos primeros 7 registros

summary(datos5)

      anio            xt              wt              ct      
 Min.   :1999   Min.   :179.2   Min.   :126.4   Min.   :1141  
 1st Qu.:2003   1st Qu.:185.6   1st Qu.:139.4   1st Qu.:1538  
 Median :2006   Median :241.2   Median :143.9   Median :2246  
 Mean   :2006   Mean   :262.6   Mean   :149.1   Mean   :2168  
 3rd Qu.:2010   3rd Qu.:320.9   3rd Qu.:152.8   3rd Qu.:2735  
 Max.   :2014   Max.   :424.7   Max.   :193.5   Max.   :3082  

regresion2<-lm(log(xt)~log(wt)+log(ct),datos5) #regresión Cobb DOuglas
summary(regresion2) #estadísticos de la regresion

Call:
lm(formula = log(xt) ~ log(wt) + log(ct), data = datos5)

Residuals:
     Min       1Q   Median       3Q      Max 
-0.16168 -0.04704 -0.01696  0.03868  0.16616 

Coefficients:
            Estimate Std. Error t value Pr(>|t|)    
(Intercept)  -7.2869     2.2966  -3.173  0.00734 ** 
log(wt)       0.9402     0.3144   2.991  0.01042 *  
log(ct)       1.0637     0.1119   9.506 3.23e-07 ***
---
Signif. codes:  0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1

Residual standard error: 0.09381 on 13 degrees of freedom
Multiple R-squared:  0.9139,    Adjusted R-squared:  0.9006 
F-statistic: 68.95 on 2 and 13 DF,  p-value: 1.2e-07

dwtest(regresion2) #si hay autocorrelación

    Durbin-Watson test

data:  regresion2
DW = 0.73868, p-value = 0.0001125
alternative hypothesis: true autocorrelation is greater than 0

cor(datos5) #si hay multicolinealidad

           anio         xt         wt         ct
anio  1.0000000  0.9362228 -0.7339226  0.9930411
xt    0.9362228  1.0000000 -0.5069428  0.9234758
wt   -0.7339226 -0.5069428  1.0000000 -0.7120275
ct    0.9930411  0.9234758 -0.7120275  1.0000000

Regresión UNAM

url2<-"https://raw.githubusercontent.com/armandovl/datasets_uno/main/CAP2_MCO_0.csv"
datos2<-read.csv(url(url2))

head(datos2)

modelo2<-lm(y~x2+x3,datos2)

summary(modelo2)

Call:
lm(formula = y ~ x2 + x3, data = datos2)

Residuals:
     Min       1Q   Median       3Q      Max 
-0.20360 -0.08483  0.01550  0.06820  0.28696 

Coefficients:
             Estimate Std. Error t value Pr(>|t|)    
(Intercept)  1.381549   0.131886  10.475 2.75e-13 ***
x2           0.022279   0.001305  17.075  < 2e-16 ***
x3          -0.003898   0.007560  -0.516    0.609    
---
Signif. codes:  0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1

Residual standard error: 0.1016 on 42 degrees of freedom
Multiple R-squared:  0.9752,    Adjusted R-squared:  0.974 
F-statistic: 825.3 on 2 and 42 DF,  p-value: < 2.2e-16

plot(modelo2)

url3<-"https://raw.githubusercontent.com/armandovl/datasets_uno/main/CAP2_MCO_0.csv"
datos2<-read.csv(url(url3))