porto.Rmd

---
title: "Untitled"
output: html_document
---

```{r setup, include=FALSE}
knitr::opts_chunk$set(echo = TRUE)

library(plyr)
library(tidyverse)
library(tidyr)
library(readr)
library(lsr)
library(caret)
library(e1071)
library(gbm)

all_data <- read_csv('train.csv')




```


```{r divide into train and test}

n <- dim(all_data)[1]

# number of observations that go in the training set
n_tr <- floor(n * .8)


# randomly select n_tr numbers, without replacement, from 1...n
set.seed(42)
tr_indices <- sample(x=1:n, size=n_tr, replace=FALSE)

# break the data into a non-overlapping train and test set
train <- all_data[tr_indices, ]
test <- all_data[-tr_indices, ]

```

```{r}
tester=gbm(target ~ . ,train[,2:59], distribution = "gaussian",n.trees = 10000, shrinkage = 0.01, interaction.depth = 4)
n.trees = seq(from=100 ,to=10000, by=100) #no of trees-a vector of 100 values 
predmatrix<-predict(tester,test[,2:59], n.trees = n.trees)
test.error<-with(test[,2:59],apply( (predmatrix-target)^2,2,mean))
plot(n.trees , test.error , pch=19,col="blue",xlab="Number of Trees",ylab="Test Error", main = "Perfomance of Boosting on Test Set")


#fitControl <- trainControl(method = "repeatedcv", number = 4, repeats = 4)
#set.seed(33)
#gbmFit1 <- train(as.factor(target) ~ ., data = train[,2:59], method = "gbm", trControl = fitControl,verbose = FALSE)

```


```{r}
reg_1 <- lm(target ~ (ps_ind_07_bin + ps_ind_17_bin + ps_reg_02 + ps_car_13 + ps_car_12 + ps_calc_03), data = train)
pred_1<- predict(reg_1, newdata = test)
test <- mutate(test,our_preds_1=pred_1)
test <- test %>% 
    mutate(ResidSqr = (target - our_preds_1)^2)


print("Mean residual squared:")
mean(test$ResidSqr)


test_final <- read_csv ('test.csv')
predictions <- data.frame(predict(reg_1, test_final))
write_csv(predictions, 'reg_predictions.csv')


```

test_final <- read_csv ('test.csv')
predictions <- data.frame(predict(reg_1, test_final))
test_final <- mutate(test_final,pred=predictions)
write_csv(test_final, 'reg_predictions.csv')


```




```{r}


y <- data.frame(train[2])
x <- data.frame (train[3:59])




corrs <- correlate(x, y)



```