HHernandez_DATA621_Discussion08

MARR 8.6

install.packages("alr3")

## package 'alr3' successfully unpacked and MD5 sums checked
## 
## The downloaded binary packages are in
##  C:\Users\humberh\AppData\Local\Temp\RtmpgxQV2m\downloaded_packages

require("alr3")

data(banknote)
summary(banknote)

##      Length           Left           Right           Bottom      
##  Min.   :213.8   Min.   :129.0   Min.   :129.0   Min.   : 7.200  
##  1st Qu.:214.6   1st Qu.:129.9   1st Qu.:129.7   1st Qu.: 8.200  
##  Median :214.9   Median :130.2   Median :130.0   Median : 9.100  
##  Mean   :214.9   Mean   :130.1   Mean   :130.0   Mean   : 9.418  
##  3rd Qu.:215.1   3rd Qu.:130.4   3rd Qu.:130.2   3rd Qu.:10.600  
##  Max.   :216.3   Max.   :131.0   Max.   :131.1   Max.   :12.700  
##       Top           Diagonal           Y      
##  Min.   : 7.70   Min.   :137.8   Min.   :0.0  
##  1st Qu.:10.10   1st Qu.:139.5   1st Qu.:0.0  
##  Median :10.60   Median :140.4   Median :0.5  
##  Mean   :10.65   Mean   :140.5   Mean   :0.5  
##  3rd Qu.:11.20   3rd Qu.:141.5   3rd Qu.:1.0  
##  Max.   :12.30   Max.   :142.4   Max.   :1.0

# Fit model 1 - Logistic Regression
log_mod <- glm(Y~Bottom+Diagonal, data = banknote, family = "binomial")

## Warning: glm.fit: algorithm did not converge

## Warning: glm.fit: fitted probabilities numerically 0 or 1 occurred

summary (log_mod)

## 
## Call:
## glm(formula = Y ~ Bottom + Diagonal, family = "binomial", data = banknote)
## 
## Deviance Residuals: 
##       Min         1Q     Median         3Q        Max  
## -5.51e-04  -2.00e-08   0.00e+00   2.00e-08   6.58e-04  
## 
## Coefficients:
##              Estimate Std. Error z value Pr(>|z|)
## (Intercept)   99422.9  5433597.5   0.018    0.985
## Bottom          688.7    37796.3   0.018    0.985
## Diagonal       -751.8    41093.2  -0.018    0.985
## 
## (Dispersion parameter for binomial family taken to be 1)
## 
##     Null deviance: 2.7726e+02  on 199  degrees of freedom
## Residual deviance: 1.0424e-06  on 197  degrees of freedom
## AIC: 6
## 
## Number of Fisher Scoring iterations: 25

y_pred <- predict(log_mod, type = "response")
table(round(y_pred), banknote$Y)

##    
##       0   1
##   0 100   0
##   1   0 100

# Perfect Logistic Fit

# Plotting the 2 predictors

plot(banknote$Diagonal, banknote$Bottom, pch = c(1, 5)[round(y_pred)+1])