Lab 3

Usando el dataset de Titanic, realizar una comparación entre diferentes enfoques para clasificación (regresión logistica, lda, knn y otros) comparar utilizando los valores del costo y de gamma.

Usar la clase de la Semana #7 como referencia.

library(e1071)
library(dplyr)
library(ISLR)
library(ggplot2)
library(rpart)
library(rpart.plot)
library(readr)
library(caret)
library(tidyverse)
titanic <- read.csv("titanic.csv")

Limpiar data

titanic_clean <- titanic%>% select(-PassengerId, 
                                  -Name,
                                  -Ticket, 
                                  -Cabin )
titanic_clean$Survived <- factor(titanic_clean$Survived)
titanic_clean$Pclass <- factor(titanic_clean$Pclass)
titanic_clean$Sex <- factor(titanic_clean$Sex)
summary(titanic_clean)

 Survived Pclass      Sex           Age       
 0:549    1:216   female:314   Min.   : 0.42  
 1:342    2:184   male  :577   1st Qu.:22.00  
          3:491                Median :28.00  
                               Mean   :29.36  
                               3rd Qu.:35.00  
                               Max.   :80.00  
     SibSp           Parch             Fare        Embarked
 Min.   :0.000   Min.   :0.0000   Min.   :  0.00    :  2   
 1st Qu.:0.000   1st Qu.:0.0000   1st Qu.:  7.91   C:168   
 Median :0.000   Median :0.0000   Median : 14.45   Q: 77   
 Mean   :0.523   Mean   :0.3816   Mean   : 32.20   S:644   
 3rd Qu.:1.000   3rd Qu.:0.0000   3rd Qu.: 31.00           
 Max.   :8.000   Max.   :6.0000   Max.   :512.33           

str(titanic_clean)

'data.frame':   891 obs. of  8 variables:
 $ Survived: Factor w/ 2 levels "0","1": 1 2 2 2 1 1 1 1 2 2 ...
 $ Pclass  : Factor w/ 3 levels "1","2","3": 3 1 3 1 3 3 1 3 3 2 ...
 $ Sex     : Factor w/ 2 levels "female","male": 2 1 1 1 2 2 2 2 1 1 ...
 $ Age     : num  22 38 26 35 35 28 54 2 27 14 ...
 $ SibSp   : int  1 1 0 1 0 0 0 3 0 1 ...
 $ Parch   : int  0 0 0 0 0 0 0 1 2 0 ...
 $ Fare    : num   ...

Regresión logística

Se creó la regresión logística encontrando que el Cut-off ideal 0.61, con una especificidad de 0.948717949 y una sensibilidad de .0.5892857

train_glm_index <- sample(1:nrow(titanic_clean),nrow(titanic_clean)*0.7)
train_glm<-titanic_clean[train_glm_index,]
test_glm <- titanic_clean[-train_glm_index,]
glm_titanic = glm(Survived~., data=train_glm, family = "binomial")
pred_prob<- predict(glm_titanic, test_glm, type="response")
roc_data<- function(){
  roc_out<- data_frame(especificidad=0,sensibilidad=0, cutoff=0)
  for(i in seq(0.01,0.95, by=0.01)){
    pred <- ifelse(pred_prob>i, 1, 0)
    cm<- table(pred, Original = test_glm$Survived)
    (sensibilidad <- cm[2,2]/sum(cm[,2]))
    (especificidad <- cm[1,1]/sum(cm[,1]))
    roc_out <-rbind(roc_out, c(especificidad, sensibilidad,i))
  }
  return(roc_out[-1,])
}
View(roc_plot_data)
plot(roc_plot_data[,1:2], type= "l")

SVM

Se calculó las SVM y se encontró que el modelo con mayor accuracy es con Gamma =1 y Cost = 2.

for (i in 1:10){
  for (j in 1:10){
    svm_titanic <-svm(Survived~., data = train, gamma = i, cost = j)
    pred <- predict(svm_titanic, test)
    cm<- table (pred, original = test$Survived)
    accuracy <- (cm[1,1]+cm[2,2])/sum(cm)
    svm_data <- cbind(gamma = i, cost = j, accuracy)
    
  }
}
for (i in 1:10){
  for (j in 1:10){
    svm_titanic <-svm(Survived~., data = train, gamma = i, cost = j)
    pred <- predict(svm_titanic, test)
    cm<- table (pred, original = test$Survived)
    accuracy <- (cm[1,1]+cm[2,2])/sum(cm)
    svm_data <- cbind(gamma = i, cost = j, accuracy)
    
  }
}