Laboratorio 3

Fiabilidad

Brayan Cruz Corona

13001595

2018

Basado en lo que se solicita en la Semana 7 y utilizando el dataset de Titanic, se procede a buscar el valor mas grande en base a gamma y

Se procede a limpiar toda la data, se almacena en la matriz titanic_clean

titanic <- read.csv("titanic.csv")
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)
titanic_clean$Embarked <- factor(titanic_clean$Embarked)
titanic_clean$Age[is.na(titanic_clean$Age)] <- median(titanic_clean$Age, na.rm = TRUE)
titanic_clean$Embarked[is.na(titanic_clean$Embarked)] <- "S"
summary(titanic_clean)

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

Separamos la data en 2 datasets, uno para entrenar “Train” y el otro para hacer las pruebas “Test”, train contiene el 70% de los datos de titanic_clear:

train_index <- sample(1:nrow(titanic_clean),nrow(titanic_clean)*0.7)
train <- titanic_clean[train_index,]
test <- titanic_clean[-train_index,]

Creamos un ciclo para averiguar la combinacion optima de gamma y cost que nos da los mejores resultados:

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

           [,1]      [,2]      [,3]      [,4]      [,5]      [,6]      [,7]      [,8]      [,9]     [,10]
 [1,] 0.7798507 0.7835821 0.7835821 0.7798507 0.7798507 0.7873134 0.7910448 0.7910448 0.7835821 0.7835821
 [2,] 0.7574627 0.7649254 0.7649254 0.7611940 0.7574627 0.7574627 0.7574627 0.7574627 0.7611940 0.7574627
 [3,] 0.7462687 0.7537313 0.7574627 0.7574627 0.7611940 0.7611940 0.7574627 0.7611940 0.7611940 0.7611940
 [4,] 0.7313433 0.7388060 0.7388060 0.7388060 0.7350746 0.7350746 0.7350746 0.7313433 0.7313433 0.7425373
 [5,] 0.7350746 0.7350746 0.7350746 0.7313433 0.7276119 0.7276119 0.7313433 0.7313433 0.7313433 0.7313433
 [6,] 0.7276119 0.7350746 0.7276119 0.7276119 0.7276119 0.7313433 0.7313433 0.7313433 0.7313433 0.7313433
 [7,] 0.7126866 0.7313433 0.7238806 0.7238806 0.7276119 0.7276119 0.7238806 0.7313433 0.7313433 0.7313433
 [8,] 0.7126866 0.7164179 0.7238806 0.7238806 0.7201493 0.7201493 0.7313433 0.7313433 0.7313433 0.7313433
 [9,] 0.7164179 0.7126866 0.7126866 0.7126866 0.7201493 0.7238806 0.7238806 0.7238806 0.7201493 0.7201493
[10,] 0.7126866 0.7164179 0.7164179 0.7126866 0.7201493 0.7238806 0.7201493 0.7201493 0.7201493 0.7201493

Por ultimo averiguamos las posiciones en donde los mejores resultados se ubican en base a gamma y cost:

which(acc_matrix == max(acc_matrix), arr.ind = T)

     row col
[1,]   1   7
[2,]   1   8