uploading the template for preprocessing

Author: algorithmica-repository

kaggle/targeted-marketing-springleaf/springleaf-v1.R

@@ -0,0 +1,99 @@
+library(caret)
+library(randomForest)
+library(ggplot2)
+library(outliers)
+
+setwd("E:/data analytics/kaggle/springleaf")
+
+//Read the dataset from train file
+trainSet = read.csv("train.csv", header = TRUE, na.strings=c("NA","","-1","-99999"))
+dim(trainSet)
+str(trainSet)
+head(trainSet)
+
+set.seed(42)
+trainSet = trainSet[sample(nrow(trainSet),20000),]
+dim(trainSet)
+
+//Remove the variables which have 95% NAs
+threshold_val = 0.95 * dim(trainSet)[1]
+include_cols = !apply(trainSet, 2, function(y) sum(is.na(y)) > threshold_val)
+trainSet = trainSet[, include_cols]
+
+//Remove the variables which have very less variance
+nearZvar = nearZeroVar(trainSet, saveMetrics = TRUE)
+trainSet = trainSet[ ,nearZvar$nzv==FALSE]
+
+//Remove the variables which are highly correlated
+corr_matrix = abs(cor(trainSet[,-dim(trainSet)[2]]))
+diag(corr_matrix) = 0
+correlated_col = findCorrelation(corr_matrix, verbose = FALSE , cutoff = .95)
+trainSet = trainSet[, -c(correlated_col)]
+
+
+
+
+
+# Set a random seed 
+
+
+#model tuning strategy
+ctrl = trainControl(method = "cv", # Use cross-validation
+                    number = 10) # Use 10 folds for cross-validation
+
+model_dt = train(Survived ~ Pclass + Sex + Embarked + Fare, 
+                 data = trainSet, 
+                 method = "rpart",
+                 trControl = ctrl)
+model_dt$finalModel
+model_dt
+
+model_dt = train(Survived ~ Pclass + Sex + Embarked + Fare, 
+                 data = trainSet, 
+                 method = "rpart",
+                 trControl = ctrl,
+                 tuneLength=10)
+model_dt$finalModel
+model_dt
+
+grid=data.frame(.cp=c(0,0.1,0.6))
+model_dt = train(Survived ~ Pclass + Sex + Embarked + Fare, 
+                 data = trainSet, 
+                 method = "rpart",
+                 trControl = ctrl,
+                 tuneGrid = grid)
+model_dt$finalModel
+model_dt
+
+model_dt = train(Survived ~ Pclass + Sex + Age + Embarked + SibSp + Fare, 
+                 data = trainSet, control=rpart.control(minsplit=2), 
+                 method = "rpart",
+                 trControl = ctrl)
+model_dt$finalModel
+model_dt
+
+model_dt = train(Survived ~ Pclass + Sex + Age + Embarked + SibSp + Fare, 
+                 data = trainSet, 
+                 method = "J48",
+                 trControl = ctrl)
+model_dt$finalModel
+
+
+testSet = read.table("test.csv", sep = ",", header = TRUE)
+dim(testSet)
+str(testSet)
+head(testSet)
+testSet$Pclass = factor(testSet$Pclass)
+summary(testSet)
+testSet$Fare = ifelse(is.na(testSet$Fare), mean(testSet$Fare, na.rm = TRUE), testSet$Fare)
+
+
+testSet$Survived = predict(model_dt, newdata = testSet)
+
+submission = testSet[,c("PassengerId", "Survived")]
+
+write.table(submission, file = "submission.csv", col.names = TRUE, row.names = FALSE, sep = ",")
+
+library()
+search()
+ls("package:base")