• This project is related to the Online fraud In finanace.
• The Objective of this Project is detect the fraud .
• The dataset contains more than 6 lakhs rows.
• Most of columns is equal to zero.
• The dataset is Imbalanced .

---

dataset link:-https://www.kaggle.com/datasets/rupakroy/online-payments-fraud-detection-dataset

WITHOUT Handling imbalanced dataset:-

---

Best Model:-

1. Decision Tree :-

Class	Precision	Recall	F1-Score	Support
0	1.00	1.00	1.00	1,876,979
1	0.88	0.84	0.86	2,443
Accuracy	-	-	1.00	1,879,422
Macro Avg	0.94	0.92	0.93	1,879,422
Weighted Avg	1.00	1.00	1.00	1,879,422

2. Random Forest :-

Class	Precision	Recall	F1-Score	Support
0	1.00	1.00	1.00	1,876,979
1	0.94	0.70	0.81	2,443
Accuracy	-	-	1.00	1,879,422
Macro Avg	0.97	0.85	0.90	1,879,422
Weighted Avg	1.00	1.00	1.00	1,879,422

---

Handling Imbalaced Dataset

---

1. Undersampling: Best Model:-

• Decision Tree:- Too Bad for precison and f1-score

Class	Precision	Recall	F1-Score	Support
0	1.00	0.99	0.99	1,876,979
1	0.09	1.00	0.17	2,443
Accuracy	-	-	0.99	1,879,422
Macro Avg	0.55	0.99	0.58	1,879,422
Weighted Avg	1.00	0.99	0.99	1,879,422

---

2. Oversampling Best Model:-

• Decision Tree :-

Class	Precision	Recall	F1-Score	Support
0	1.00	1.00	1.00	1,876,979
1	0.86	0.80	0.83	2,443
Accuracy	-	-	1.00	1,879,422
Macro Avg	0.93	0.90	0.92	1,879,422
Weighted Avg	1.00	1.00	1.00	1,879,422

---

3. SMOTE Best Model:-

• Decision Tree :-

Class	Precision	Recall	F1-Score	Support
0	1.00	1.00	1.00	1,876,979
1	0.61	0.97	0.75	2,443
Accuracy	-	-	1.00	1,879,422
Macro Avg	0.81	0.99	0.88	1,879,422
Weighted Avg	1.00	1.00	1.00	1,879,422

---

4. Ensemble technique:

• BalancedRandomForestClassifier: Precision and F1-score is too bad

Class	Precision	Recall	F1-Score	Support
0	1.00	0.98	0.99	1,876,979
1	0.07	1.00	0.13	2,443
Accuracy	-	-	0.98	1,879,422
Macro Avg	0.53	0.99	0.56	1,879,422
Weighted Avg	1.00	0.98	0.99	1,879,422

CONCLUSION

• The decision tree is working well for all cases . We can use decision tree models for predictions.