Chapter_1_Example.Rmd

---
title: "Chapter_1_Example"
output: github_document
---

```{r build model}
library("rpart")

load("GCDData.RData")

model <- rpart(Good_Loan ~ 
                 Duration_in_month +
                 Installment_rate_in_percentage_of_disposable_income +
                 Credit_amount +
                 Other_installment_plans,
               data = d,
               control = rpart.control(maxdept = 4),
               method = "class")
```

```{r present_model}
library("rpart.plot")

print(model)

rpart.plot(model)
```

```{r add_column}
d$Loan.status <- d$Good_Loan
```


```{r example_1.1_of_section_1.2.4 }
# example 1.1 of section 1.2.4 
# (example 1.1 of section 1.2.4)  : The data science process : Stages of a data science project : Model evaluation and critique 
# Title: Calculating the confusion matrix 

conf_mat <- table(actual = d$Loan.status, pred = predict(model, type = 'class'))  	# Note: 1 
conf_mat


(accuracy <- sum(diag(conf_mat)) / sum(conf_mat))   	# Note: 2 

(precision <- conf_mat["BadLoan", "BadLoan"] / sum(conf_mat[, "BadLoan"])) 	# Note: 3 

(recall <- conf_mat["BadLoan", "BadLoan"] / sum(conf_mat["BadLoan", ])) 	# Note: 4 

(fpr <- conf_mat["GoodLoan","BadLoan"] / sum(conf_mat["GoodLoan", ])) 	# Note: 5 


# Note 1: 
#   Create the confusion matrix. 

# Note 2: 
#   accuracy 
#   confusion matrix 
#   Overall model accuracy: 73% of the 
#   predictions were correct. 

# Note 3: 
#   precision 
#   confusion matrix 
#   Model precision: 76% of the 
#   applicants predicted as bad really did 
#   default. 

# Note 4: 
#   recall 
#   confusion matrix 
#   Model recall: the model found 14% of 
#   the defaulting loans. 

# Note 5: 
#   false positive rate 
#   confusion matrix 
#   False positive rate: 2% of the good 
#   applicants were mistakenly identified as 
#   bad. 
```