Update randomForest.R

randomForest.R

@@ -1,5 +1,4 @@
 
-
 setwd("~/PycharmProjects/titanic")
 
 
@@ -25,7 +24,8 @@ exp<- t[,c(3,5:8,10:12)]
 colnames(exp)
 
 x<- data.frame(target,exp)
-x<-x[sapply(x, function(x) !any(is.na(x)))] 
+#x<-x[sapply(x, function(x) !any(is.na(x)))] 
+#vamos a atender el asunto de los NA 
 
 str(x)
 
@@ -34,141 +34,8 @@ str(x)
 x$Sex<-ifelse(x$Sex=="female",0,1)
 CABIN<- lapply(x$Cabin, substring, 1, 1)
 x$Cabin<-unlist(CABIN)
-table(x$Embarked)
-empty <- x$Embarked == ""
-x$Embarked[empty] <- "S"
-x$Embarked<-ifelse(x$Embarked=="C",1,ifelse(x$Embarked=="Q",2,3))
-x$Cabin<-as.factor(x$Cabin)
-table(x$Cabin)
-x$Cabin<-ifelse(x$Cabin=="A",1,ifelse(x$Cabin=="B",
-                                      2,ifelse(x$Cabin=="C",3,
-                                               ifelse(x$Cabin=="D",
-                                                      4,ifelse(x$Cabin=="E",
-                                                               5,ifelse(x$Cabin=="F",
-                                                                        6,ifelse(x$Cabin=="G",
-                                                                                 7,ifelse(x$Cabin=="T",8,9))))))))
-
-
-
-str(x)
-
-
-x<-lapply(x, normalize)
-
-x<-as.data.frame(x)
-
-set.seed(7)
-
-rand <- sample(1:nrow(x), nrow(x))
-
-# train is our training sample.
-train = x[rand[1:600], ]
-
-# Create a holdout set for evaluating model performance.
-# Note: cross-validation is even better than a single holdout sample.
-
-cv = x[601:nrow(x), ]
-
-# Review the outcome variable distribution.CLASIFICACIOON
-table(Y_train, useNA = "ifany")
-
-# Set the seed for reproducibility.
-set.seed(1)
-
-RF<-randomForest(as.factor(train$target)~.,
-                 data = train[,-1],importance=TRUE,proximity=T,ntree=500)
-
-p<-predict(RF,cv[,-1])
-
-Accuracy(p,cv$target)
-
-########################### intentemos otros modelos que tengo a mano 
-
-set.seed(1)
-
-RF<-svm(as.factor(train$target)~.,data = train[,-1],
-        scale = TRUE,kernel="polynomial",type="C-classification",degree=3)
-
-p<-predict(RF,cv[,-1])
-
-Accuracy(p,cv$target)
 
 
-#################################################
-
-model <- keras_model_sequential()
-
-model %>%
-  layer_dense(units = ncol(train[,-1]), activation = 'relu',
-              input_shape = ncol(train[,-1]), ) %>%
-  layer_dropout(rate = 0.99) %>%
-  layer_dense(units = 10, activation = 'relu') %>%
-  layer_dropout(rate = 0.95) %>%
-  layer_dense(units = 1, activation = 'sigmoid')
-
-history <- model %>% compile(
-  loss = 'binary_crossentropy',
-  optimizer = optimizer_adam(lr=0.001),
-  metrics = c('accuracy')
-)
-
-model %>% fit(
-  as.matrix(train[,-1],dimname=NULL), train$target,
-  epochs =5,
-  batch_size = 20,
-  validation_split = 0.0
-)
-
-#####################################################
-
-#intentemos sacar mas informacion de la data
-
-
-View(t$Name)
-View(t$Ticket)
-TT<- lapply(t$Ticket, substring, 1, 1)
-x$Ticket<-as.factor(unlist(TT))
-table(x$Ticket)
-
-
-
-#################con la data nueva
-
-setwd("~/PycharmProjects/titanic")
-
-
-library(caret)
-library(randomForest)
-library(MLmetrics)
-library(e1071)
-library (keras)
-
-
-normalize <- function(x) { 
-  return((x - min(x)) / (max(x) - min(x)))
-}
-
-
-# read data.frame with target and paste it to explanatory variables data.frame
-
-t <- read.csv("train.csv")
-target <- t$Survived
-
-colnames(t)
-exp<- t[,c(3,5:8,10:12)]
-colnames(exp)
-
-x<- data.frame(target,exp)
-x<-x[sapply(x, function(x) !any(is.na(x)))] 
-
-str(x)
-
-#tenemos tres variables con factores, vamos a cambiarlas a numericas
-
-x$Sex<-ifelse(x$Sex=="female",0,1)
-CABIN<- lapply(x$Cabin, substring, 1, 1)
-x$Cabin<-unlist(CABIN)
-
 
 table(x$Embarked)
 empty <- x$Embarked == ""
@@ -184,11 +51,10 @@ x$Cabin<-ifelse(x$Cabin=="A",1,ifelse(x$Cabin=="B",
                                                                         6,ifelse(x$Cabin=="G",
                                                                                  7,ifelse(x$Cabin=="T",8,9))))))))
 
-
+x<-na.aggregate(x)
 
 str(x)
 
-
 x<-lapply(x, normalize)
 
 x<-as.data.frame(x)
@@ -224,7 +90,3 @@ Accuracy(p,cv$target)
 
 
 
-
-
-
-