Cost Comparsion.Rmd

---
title: "Cost and Comparison"
author: "Angel"
date: "2025-12-03"
output: html_document
---

```{r setup, include=FALSE}
knitr::opts_chunk$set(echo = TRUE)
library(ggplot2)
```

## Cost Calculation 

The cost of contacting a customer and the revenue after client accpeting the campaign are given by the dataset which we will use. 

The cost to contact: $30
The revenue after client accept campaign: $110 

We will be using the sensitivity and specificity to calculate the potential revenue earn an individual customer. 


```{r}
cost <- 30
revenue <- 110

extract_metrics_cost <- function(cm) {
  sensitivity <- as.numeric(cm$byClass['Sensitivity'])
  specificity <- as.numeric(cm$byClass['Specificity'])
  
  expected_profit <- (revenue - cost) * sensitivity -
                      cost * (1-specificity)
  data.frame(
    Expected_profit = expected_profit
  )
}
comparison_table <- rbind(
  GLM = extract_metrics_cost(lr_cm_interactions),
  KNN = extract_metrics_cost(knn_advanced_cm),
  ANN = extract_metrics_cost(ann_base_matrix),
  DecisionTree = extract_metrics_cost(dt_cm),
  SVM = extract_metrics_cost(svm_cm),
  RandomForest = extract_metrics_cost(rf_cm),
  CostStack = extract_metrics_cost(cost_cm)
)
kable(comparison_table, 
      caption = "Comparison of All Models: Expected Profit",
      digits = 2) %>%
  kable_styling(bootstrap_options = c("striped", "hover", "condensed", "responsive"),
                full_width = F,
                position = "center") %>%
  row_spec(0, bold = TRUE, color = "white", background = "black")
```

Cost Combine Decision Tree Result 
```{r}
library(ggplot2)
library(dplyr)

# Convert confusion matrix to data frame
cm_to_df <- function(cm) {
  cm_table <- as.data.frame(cm$table)
  total <- sum(cm_table$Freq)
  
  cm_table <- cm_table %>%
    mutate(Percentage = round((Freq / total) * 100, 1),
           Label = paste(Freq, "\n(", Percentage, "%)", sep=""))
  
  return(cm_table)
}

# Create plot
plot_confusion_matrix <- function(cm) {
  cm_df <- cm_to_df(cm)
  
  # Reverse the factor levels for Prediction to match your desired display
  cm_df$Reference <- factor(cm_df$Reference, levels = c("0", "1"))
  cm_df$Prediction <- factor(cm_df$Prediction, levels = c("1", "0"))  # REVERSED!
  
  ggplot(data = cm_df, aes(x = Reference, y = Prediction)) +
    geom_tile(aes(fill = Freq), color = "black") +
    geom_text(aes(label = Label), size = 6, color = "white") +
    scale_fill_gradient(low = "lightblue", high = "blue") +
    scale_x_discrete(limits = c("0", "1")) +
    # No scale_y_discrete needed since factor levels already set correctly
    labs(title = "Result Using Cost Combined Decision Tree Model",
         x = "Reference",
         y = "Prediction") +
    theme_minimal() +
    theme(axis.text = element_text(size = 12),
          axis.title = element_text(size = 14),
          plot.title = element_text(size = 16, hjust = 0.5))
}

# Generate the plot
plot_confusion_matrix(cost_cm)
```