Decision tree Model : Output Summary
loans_model_HK
##
## Model formula:
## default ~ fixed_assets + months_loan + credit_record + purpose +
## amount + savings_balance + employment + age + other_credit +
## housing + loans_count + job + dependents
##
## Fitted party:
## [1] root
## | [2] months_loan <= 33
## | | [3] housing in other, rent
## | | | [4] credit_record in critical, poor: no (n = 122, err = 23.8%)
## | | | [5] credit_record in good, perfect
## | | | | [6] credit_record in good: no (n = 229, err = 45.9%)
## | | | | [7] credit_record in perfect
## | | | | | [8] age <= 26: yes (n = 9, err = 0.0%)
## | | | | | [9] age > 26: yes (n = 11, err = 27.3%)
## | | [10] housing in own
## | | | [11] employment < 1 year, unemployed
## | | | | [12] savings_balance < 0.1M HKD, 0.1 - 0.5M HKD: no (n = 165, err = 46.7%)
## | | | | [13] savings_balance > 1000 HKD, 0.5 - 1M HKD, unknown
## | | | | | [14] amount <= 4716: no (n = 33, err = 6.1%)
## | | | | | [15] amount > 4716: no (n = 10, err = 40.0%)
## | | | [16] employment > 10 years, 1 - 3 years, 3 - 10 years: no (n = 638, err = 21.9%)
## | [17] months_loan > 33
## | | [18] housing in other, rent
## | | | [19] loans_count <= 1: yes (n = 62, err = 40.3%)
## | | | [20] loans_count > 1: yes (n = 34, err = 8.8%)
## | | [21] housing in own: no (n = 187, err = 43.3%)
##
## Number of inner nodes: 10
## Number of terminal nodes: 11
Decision tree Model : Confusion Matrix
loans_pred_HK = predict(loans_model_HK, loans_test)
HK_conft = table("prediction" = loans_pred_HK, "actual" = loans_test$default)
confusionMatrix(loans_pred_HK, loans_test$default, positive = "yes")
## Confusion Matrix and Statistics
##
## Reference
## Prediction no yes
## no 323 95
## yes 7 75
##
## Accuracy : 0.796
## 95% CI : (0.758, 0.8305)
## No Information Rate : 0.66
## P-Value [Acc > NIR] : 1.522e-11
##
## Kappa : 0.4802
##
## Mcnemar's Test P-Value : < 2.2e-16
##
## Sensitivity : 0.4412
## Specificity : 0.9788
## Pos Pred Value : 0.9146
## Neg Pred Value : 0.7727
## Prevalence : 0.3400
## Detection Rate : 0.1500
## Detection Prevalence : 0.1640
## Balanced Accuracy : 0.7100
##
## 'Positive' Class : yes
##
Confusion Matrix: Test Data: Credit Default
|
Predicted
|
Actual Class
|
|
|
no
|
yes
|
|
no
|
323
|
95
|
|
yes
|
7
|
75
|
