Ssssss— title: “Untitled” author: “J” date: “15 December 2018” output: html_document —

Modelo general

crelogit <- glm(credito ~ ahorro + educ + venta + cultivo + sierra +
                  edad + produccion + sexo + provincia, 
                family = binomial (link = "logit"),
                data = pu)
coeftest(crelogit, vcov = vcovHC, tipe = "HC1")

## 
## z test of coefficients:
## 
##                Estimate  Std. Error z value  Pr(>|z|)    
## (Intercept) -1.8652e+00  3.5868e-01 -5.2003 1.989e-07 ***
## ahorro       4.5628e-01  1.9205e-01  2.3758 0.0175111 *  
## educ         2.1176e-01  3.1023e-02  6.8259 8.740e-12 ***
## venta        6.2252e-05  1.2995e-05  4.7904 1.665e-06 ***
## cultivo      6.2483e-02  2.0085e-01  0.3111 0.7557301    
## sierra       4.9423e-03  2.3952e-01  0.0206 0.9835375    
## edad        -3.3139e-03  4.1545e-03 -0.7977 0.4250711    
## produccion   2.0684e-05  6.6327e-05  0.3118 0.7551555    
## sexo         5.6020e-02  1.5747e-01  0.3557 0.7220308    
## provincia   -1.2923e-01  3.7635e-02 -3.4338 0.0005953 ***
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1

el modelo corregido

crelogit1 <- glm(credito ~ ahorro + educ + venta  + provincia, 
                family = binomial (link = "logit"), 
                data = pu)
coeftest(crelogit1, vcov = vcovHC, tipe = "HC1")

## 
## z test of coefficients:
## 
##                Estimate  Std. Error z value  Pr(>|z|)    
## (Intercept) -1.9505e+00  2.7574e-01 -7.0735 1.511e-12 ***
## ahorro       4.4859e-01  1.9107e-01  2.3478 0.0188846 *  
## educ         2.1846e-01  2.9553e-02  7.3923 1.443e-13 ***
## venta        6.3228e-05  1.2643e-05  5.0010 5.703e-07 ***
## provincia   -1.3063e-01  3.6915e-02 -3.5386 0.0004022 ***
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1

summary(crelogit1)$coefficients[,1]

##   (Intercept)        ahorro          educ         venta     provincia 
## -1.950451e+00  4.485901e-01  2.184627e-01  6.322776e-05 -1.306295e-01

z<-summary(crelogit1)$coefficients[,4]
p <- (1 - pnorm(abs(z), 0, 1)) * 2
1-p

##  (Intercept)       ahorro         educ        venta    provincia 
## 1.598721e-14 1.057290e-02 6.177281e-13 1.088019e-14 2.448599e-04

#intervalos de confianza
confint(crelogit1)

## Waiting for profiling to be done...

##                     2.5 %        97.5 %
## (Intercept) -2.458451e+00 -1.458141e+00
## ahorro       8.974062e-02  8.006228e-01
## educ         1.589908e-01  2.786228e-01
## venta        4.780157e-05  7.999766e-05
## provincia   -2.018987e-01 -5.988234e-02

coefplot(crelogit1)+theme_minimal() 

allEffects(crelogit1)

##  model: credito ~ ahorro + educ + venta + provincia
## 
##  ahorro effect
## ahorro
##         0       0.2       0.5       0.8         1 
## 0.1991556 0.2138518 0.2373461 0.2625595 0.2802968 
## 
##  educ effect
## educ
##          0          2          5          8         10 
## 0.09595075 0.14110731 0.24035482 0.37863475 0.48539875 
## 
##  venta effect
## venta
##         0     40000     90000     1e+05     2e+05 
## 0.1561899 0.6989400 0.9820780 0.9903960 0.9999826 
## 
##  provincia effect
## provincia
##         1         3         5         7         9 
## 0.3436086 0.2873042 0.2368962 0.1929381 0.1554749

plot(allEffects(crelogit1))

effectos marginales

round(mfx::logitmfx(credito ~ ahorro + educ + venta +  provincia, pu)$mfxest,6)

##               dF/dx Std. Err.         z    P>|z|
## ahorro     0.081141  0.034860  2.327619 0.019932
## educ       0.036853  0.005006  7.361222 0.000000
## venta      0.000011  0.000001  7.150001 0.000000
## provincia -0.022036  0.005979 -3.685349 0.000228