Guess Algoritm

#1. Guess Algoritm (Variavéis resposta: Categorica variavel explicativa: Numerica (1)) - Só importa a variavel explicativa neste algoritmo

Algoritmo que escolhe p% de vezes um resultado. O treino deste algoritmo prende-se em otimizar p, isto é, o valor de p que retorna maior accuracy nos testes.

##1.Exportação de Dados

# define the outcome and predictors
data(heights)
y <- heights$sex
x <- heights$height
#Divisão do Treino e do Teste
set.seed(2)
test_index <- createDataPartition(y, times = 1, p = 0.5, list = FALSE)
test_set <- heights[test_index, ]
train_set <- heights[-test_index, ]
head(test_set)

##       sex height
## 3    Male     68
## 4    Male     74
## 5    Male     61
## 6  Female     65
## 8  Female     62
## 13   Male     69

##2.Analise de dados

ggpairs(heights, columns = 1:2, ggplot2::aes(colour=sex))

## `stat_bin()` using `bins = 30`. Pick better value with `binwidth`.

##3.Guess Algoritm

# guess the outcome
pp <- seq(0, 1, 0.001)
accuracy <- map_df(pp, function(p){   ##Este algoritmo descobre o melhor p 
  n <- length(test_index)
  y_hat <- sample(c("Male", "Female"), n, replace = TRUE, prob=c(p, 1-p)) %>% 
  factor(levels = levels(train_set$sex))
  accu=mean(y_hat == train_set$sex)
  tibble(accura=accu)
})
p=pp[which.max(accuracy$accura)]
p

## [1] 0.999

n <- length(test_index)
y_hat <- sample(c("Male", "Female"), n, replace = TRUE, prob=c(p, 1-p)) %>% 
  factor(levels = levels(test_set$sex))
mean(y_hat == test_set$sex)

## [1] 0.7733333

confusionMatrix(data = y_hat,reference = test_set$sex)

## Confusion Matrix and Statistics
## 
##           Reference
## Prediction Female Male
##     Female      0    0
##     Male      119  406
##                                           
##                Accuracy : 0.7733          
##                  95% CI : (0.7351, 0.8085)
##     No Information Rate : 0.7733          
##     P-Value [Acc > NIR] : 0.5246          
##                                           
##                   Kappa : 0               
##                                           
##  Mcnemar's Test P-Value : <2e-16          
##                                           
##             Sensitivity : 0.0000          
##             Specificity : 1.0000          
##          Pos Pred Value :    NaN          
##          Neg Pred Value : 0.7733          
##              Prevalence : 0.2267          
##          Detection Rate : 0.0000          
##    Detection Prevalence : 0.0000          
##       Balanced Accuracy : 0.5000          
##                                           
##        'Positive' Class : Female          
##