Credit risk analysis using ML models!

Loading the needed packages

library(tidyverse)
library(tidymodels)
library(janitor)
library(themis)
library(vip)
library(rpart.plot)

Loading the datset

risk <- read_csv("~/Desktop/R/csv/german_credit.csv") %>% clean_names()
df <- risk %>% slice_head(n = 1000)
new_data <- risk %>% slice_tail(n = 10) %>% select(!creditability)

EDA

Plotting the nominal (target) variable

Plotting the numerical (predictor) variables

MODELLING

1.Logistic Regression model

Predicting on new data

## # A tibble: 10 × 3
##    .pred_class .pred_Yes .pred_No
##    <fct>           <dbl>    <dbl>
##  1 No             0.154    0.846 
##  2 No             0.162    0.838 
##  3 Yes            0.955    0.0452
##  4 No             0.332    0.668 
##  5 No             0.0986   0.901 
##  6 Yes            0.608    0.392 
##  7 No             0.484    0.516 
##  8 Yes            0.926    0.0739
##  9 No             0.441    0.559 
## 10 Yes            0.646    0.354

2.Random Forest model

Predicting on new data

## # A tibble: 10 × 3
##    .pred_class .pred_Yes .pred_No
##    <fct>           <dbl>    <dbl>
##  1 Yes             0.527    0.473
##  2 No              0.405    0.595
##  3 Yes             0.596    0.404
##  4 Yes             0.617    0.383
##  5 No              0.270    0.730
##  6 No              0.487    0.513
##  7 Yes             0.606    0.394
##  8 Yes             0.643    0.357
##  9 Yes             0.616    0.384
## 10 Yes             0.560    0.440

3.Decision Trees model

Predicting on new data

## # A tibble: 10 × 3
##    .pred_class .pred_Yes .pred_No
##    <fct>           <dbl>    <dbl>
##  1 No              0.35     0.65 
##  2 No              0.12     0.88 
##  3 No              0.444    0.556
##  4 Yes             0.786    0.214
##  5 No              0.12     0.88 
##  6 No              0.35     0.65 
##  7 Yes             1        0    
##  8 Yes             0.847    0.153
##  9 No              0.136    0.864
## 10 No              0.444    0.556

4.Comparing results

5.Confusion matrices

6.ROC curves

7.Variable Importance

8.Plotting the Decision Tree