6.Performance comparison of models trained with Decision Trees and Linear Regression methods.Rmd

---
output:
  word_document: default
  html_document: default
---
```{r}
data <- carData::Salaries
head(data)
str(data)
summary(data)
```

DATA SPLITING

```{r}
library(caret)
set.seed(2380)
index <- createDataPartition(data$salary, p = 0.75,
list = FALSE, times = 1)
train <- data[index,]
test <- data[-index,]
nrow(train)
nrow(test)
```


LINEER REGRESSION MODEL


```{r}
#Traing Model
model <- lm(salary~.,data=train)
#Performance for Train
predicted_train <- predict(model,train)
modelEvaluation_train <- data.frame(train$salary,predicted_train)
colnames(modelEvaluation_train) <- c("Actual","Predicted_train")
mse_train <- mean((modelEvaluation_train$Actual - modelEvaluation_train$Predicted_train)^2)
print(c("Train:",sqrt(mse_train)))
#Performance for Test
predicted_test <- predict(model,test)
modelEvaluation_test <- data.frame(test$salary,predicted_test)
colnames(modelEvaluation_test) <- c("Actual","Predicted_test")
mse_test <- mean((modelEvaluation_test$Actual - modelEvaluation_test$Predicted_test)^2)
print(c("Test:",sqrt(mse_test)))
```

DECISION TREE

```{r}
library(rpart)
library(rpart.plot)
#Training Model
model_dt <- rpart(salary ~. , method = "anova", data = train)
rpart.plot(model_dt)
```

```{r}
#Performance on Test
preds_test <- predict(model_dt, test)
rmse_test <- sqrt(mean((preds_test - test$salary) ^ 2))
cat("test_rmse:", rmse_test,"\n")
#Performance on Train
preds_train <- predict(model_dt, train)
rmse_train <- sqrt(mean((preds_train - train$salary) ^ 2))
cat("train_mse:", rmse_train)
```


COMPARISON
```{r}
#Comparing models performances
data.frame(
  "LRM" = c(sqrt(mse_train),sqrt(mse_test)),
  "Decision Tree" = c(rmse_train,rmse_test),
  row.names = c("Train Root Mean S. Error", "Test Root Mean S. Error")
)
```